This action might not be possible to undo. Are you sure you want to continue?
MACHINE PERCEPTION
CZECH TECHNICAL
UNIVERSITY
R
E
S
E
A
R
C
H
R
E
P
O
R
T
I
S
S
N
1
2
1
3

2
3
6
5
Statistical Pattern Recognition
Toolbox for Matlab
User’s guide
Vojtˇech Franc and V´aclav Hlav´ aˇc
¦xfrancv,hlavac¦@cmp.felk.cvut.cz
CTU–CMP–2004–08
June 24, 2004
Available at
ftp://cmp.felk.cvut.cz/pub/cmp/articles/FrancTR200408.pdf
The authors were supported by the European Union under project
IST200135454 ECVision, the project IST200132184 ActIPret and
by the Czech Grant Agency under project GACR 102/03/0440 and
by the Austrian Ministry of Education under project CONEX GZ
45.535.
Research Reports of CMP, Czech Technical University in Prague, No. 8, 2004
Published by
Center for Machine Perception, Department of Cybernetics
Faculty of Electrical Engineering, Czech Technical University
Technick´a 2, 166 27 Prague 6, Czech Republic
fax +420 2 2435 7385, phone +420 2 2435 7637, www: http://cmp.felk.cvut.cz
Statistical Pattern Recognition Toolbox for Matlab
Vojtˇech Franc and V´aclav Hlav´ aˇc
June 24, 2004
Contents
1 Introduction 3
1.1 What is the Statistical Pattern Recognition Toolbox and how it has been
developed? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 The STPRtool purpose and its potential users . . . . . . . . . . . . . . 3
1.3 A digest of the implemented methods . . . . . . . . . . . . . . . . . . . 4
1.4 Relevant Matlab toolboxes of others . . . . . . . . . . . . . . . . . . . . 5
1.5 STPRtool document organization . . . . . . . . . . . . . . . . . . . . . 5
1.6 How to contribute to STPRtool? . . . . . . . . . . . . . . . . . . . . . 5
1.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Linear Discriminant Function 8
2.1 Linear classiﬁer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Linear separation of ﬁnite sets of vectors . . . . . . . . . . . . . . . . . 10
2.2.1 Perceptron algorithm . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Kozinec’s algorithm . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Multiclass linear classiﬁer . . . . . . . . . . . . . . . . . . . . . 14
2.3 Fisher Linear Discriminant . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Generalized Anderson’s task . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.1 Original Anderson’s algorithm . . . . . . . . . . . . . . . . . . . 19
2.4.2 General algorithm framework . . . . . . . . . . . . . . . . . . . 21
2.4.3 εsolution using the Kozinec’s algorithm . . . . . . . . . . . . . 21
2.4.4 Generalized gradient optimization . . . . . . . . . . . . . . . . . 22
3 Feature extraction 25
3.1 Principal Component Analysis . . . . . . . . . . . . . . . . . . . . . . . 26
3.2 Linear Discriminant Analysis . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Kernel Principal Component Analysis . . . . . . . . . . . . . . . . . . . 31
3.4 Greedy kernel PCA algorithm . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 Generalized Discriminant Analysis . . . . . . . . . . . . . . . . . . . . . 33
3.6 Combining feature extraction and linear classiﬁer . . . . . . . . . . . . 34
1
4 Density estimation and clustering 41
4.1 Gaussian distribution and Gaussian mixture model . . . . . . . . . . . 41
4.2 MaximumLikelihood estimation of GMM . . . . . . . . . . . . . . . . 43
4.2.1 Complete data . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2.2 Incomplete data . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3 Minimax estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.4 Probabilistic output of classiﬁer . . . . . . . . . . . . . . . . . . . . . . 48
4.5 Kmeans clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5 Support vector and other kernel machines 57
5.1 Support vector classiﬁers . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 Binary Support Vector Machines . . . . . . . . . . . . . . . . . . . . . 58
5.3 OneAgainstAll decomposition for SVM . . . . . . . . . . . . . . . . . 63
5.4 OneAgainstOne decomposition for SVM . . . . . . . . . . . . . . . . . 64
5.5 Multiclass BSVM formulation . . . . . . . . . . . . . . . . . . . . . . . 66
5.6 Kernel Fisher Discriminant . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.7 Preimage problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.8 Reduced set method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6 Miscellaneous 76
6.1 Data sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.2 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.3 Bayesian classiﬁer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.4 Quadratic classiﬁer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.5 Knearest neighbors classiﬁer . . . . . . . . . . . . . . . . . . . . . . . 82
7 Examples of applications 85
7.1 Optical Character Recognition system . . . . . . . . . . . . . . . . . . 85
7.2 Image denoising . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
8 Conclusions and future planes 94
2
Chapter 1
Introduction
1.1 What is the Statistical Pattern Recognition Tool
box and how it has been developed?
The Statistical Pattern Recognition Toolbox (abbreviated STPRtool) is a collection of
pattern recognition (PR) methods implemented in Matlab.
The core of the STPRtool is comprised of statistical PR algorithms described in
the monograph Schlesinger, M.I., Hlav´ aˇc, V: Ten Lectures on Statistical and Structural
Pattern Recognition, Kluwer Academic Publishers, 2002 [26]. The STPRtool has been
developed since 1999. The ﬁrst version was a result of a diploma project of Vojtˇech
Franc [8]. Its author became a PhD student in CMP and has been extending the
STPRtool since. Currently, the STPRtool contains much wider collection of algorithms.
The STPRtool is available at
http://cmp.felk.cvut.cz/cmp/cmp software.html.
We tried to create concise and complete documentation of the STPRtool. This
document and associated help ﬁles in .html should give all information and parameters
the user needs when she/he likes to call appropriate method (typically a Matlab m
ﬁle). The intention was not to write another textbook. If the user does not know
the method then she/he will likely have to read the description of the method in a
referenced original paper or in a textbook.
1.2 The STPRtool purpose and its potential users
• The principal purpose of STPRtool is to provide basic statistical pattern recog
nition methods to (a) researchers, (b) teachers, and (c) students. The STPRtool
is likely to help both in research and teaching. It can be found useful by practi
tioners who design PR applications too.
3
• The STPRtool oﬀers a collection of the basic and the stateofthe art statistical
PR methods.
• The STPRtool contains demonstration programs, working examples and visual
ization tools which help to understand the PR methods. These teaching aids are
intended for students of the PR courses and their teachers who can easily create
examples and incorporate them into their lectures.
• The STPRtool is also intended to serve a designer of a new PR method by
providing tools for evaluation and comparison of PR methods. Many standard
and stateofthe art methods have been implemented in STPRtool.
• The STPRtool is open to contributions by others. If you develop a method
which you consider useful for the community then contact us. We are willing to
incorporate the method into the toolbox. We intend to moderate this process to
keep the toolbox consistent.
1.3 A digest of the implemented methods
This section should give the reader a quick overview of the methods implemented in
STPRtool.
• Analysis of linear discriminant function: Perceptron algorithm and multi
class modiﬁcation. Kozinec’s algorithm. Fisher Linear Discriminant. A collection
of known algorithms solving the Generalized Anderson’s Task.
• Feature extraction: Linear Discriminant Analysis. Principal Component Anal
ysis (PCA). Kernel PCA. Greedy Kernel PCA. Generalized Discriminant Analy
sis.
• Probability distribution estimation and clustering: Gaussian Mixture
Models. ExpectationMaximization algorithm. Minimax probability estimation.
Kmeans clustering.
• Support Vector and other Kernel Machines: Sequential Minimal Opti
mizer (SMO). Matlab Optimization toolbox based algorithms. Interface to the
SV M
light
software. Decomposition approaches to train the Multiclass SVM clas
siﬁers. Multiclass BSVM formulation trained by Kozinec’s algorithm, Mitchell
DemyanovMolozenov algorithm and Nearest Point Algorithm. Kernel Fisher
Discriminant.
4
1.4 Relevant Matlab toolboxes of others
Some researchers in statistical PR provide code of their methods. The Internet oﬀers
many software resources for Matlab. However, there are not many compact and well
documented packages comprised of PR methods. We found two useful packages which
are brieﬂy introduced below.
NETLAB is focused to the Neural Networks applied to PR problems. The classical
PR methods are included as well. This Matlab toolbox is closely related to the
popular PR textbook [3]. The documentation published in book [21] contains de
tailed description of the implemented methods and many working examples. The
toolbox includes: Probabilistic Principal Component Analysis, Generative Topo
graphic Mapping, methods based on the Bayesian inference, Gaussian processes,
Radial Basis Functions (RBF) networks and many others.
The toolbox can be downloaded from
http://www.ncrg.aston.ac.uk/netlab/.
PRTools is a Matlab Toolbox for Pattern Recognition [7]. Its implementation is
based on the object oriented programming principles supported by the Matlab
language. The toolbox contains a wide range of PR methods including: analysis
of linear and nonlinear classiﬁers, feature extraction, feature selection, methods
for combining classiﬁers and methods for clustering.
The toolbox can be downloaded from
http://www.ph.tn.tudelft.nl/prtools/.
1.5 STPRtool document organization
The document is divided into chapters describing the implemented methods. The last
chapter is devoted to more complex examples. Most chapters begin with deﬁnition of
the models to be analyzed (e.g., linear classiﬁer, probability densities, etc.) and the list
of the implemented methods. Individual sections have the following ﬁxed structure: (i)
deﬁnition of the problem to be solved, (ii) brief description and list of the implemented
methods, (iii) reference to the literature where the problem is treaded in details and
(iv) an example demonstrating how to solve the problem using the STPRtool.
1.6 How to contribute to STPRtool?
If you have an implementation which is consistent with the STPRtool spirit and you
want to put it into public domain through the STPRtool channel then send an email
to xfrancv@cmp.felk.cvut.cz.
5
Please prepare your code according to the pattern you ﬁnd in STPRtool implemen
tations. Provide piece of documentation too which will allow us to incorporate it to
STPRtool easily.
Of course, your method will be bound to your name in STPRtool.
1.7 Acknowledgements
First, we like to thank to Prof. M.I. Schlesinger from the Ukrainian Academy of
Sciences, Kiev, Ukraine who showed us the beauty of pattern recognition and gave us
many valuable advices.
Second, we like to thank for discussions to our colleagues in the Center for Machine
Perception, the Czech Technical University in Prague, our visitors, other STPRtool
users and students who used the toolbox in their laboratory exercises.
Third, thanks go to several Czech, EU and industrial grants which indirectly con
tributed to STPRtool development.
Forth, the STPRtool was rewritten and its documentation radically improved in
May and June 2004 with the ﬁnancial help of ECVision European Research Network
on Cognitive Computer Vision Systems (IST200135454) lead by D. Vernon.
6
Notation
Uppercase bold letters denote matrices, for instance A. Vectors are implicitly column
vectors. Vectors are denoted by lowercase bold italic letters, for instance x. The
concatenation of column vectors w ∈ R
n
, z ∈ R
m
is denoted as v = [w; z] where the
column vector v is (n + m)dimensional.
¸x x
) Dot product between vectors x and x
.
Σ Covariance matrix.
µ Mean vector (mathematical expectation).
S Scatter matrix.
A ⊆ R
n
ndimensional input space (space of observations).
¸ Set of c hidden states – class labels ¸ = ¦1, . . . , c¦.
T Feature space.
φ: A → T Mapping from input space A to the feature space T.
k: A A →R Kernel function (Mercer kernel).
T
XY
Labeled training set (more precisely multiset) T
XY
=
¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦.
T
X
Unlabeled training set T
X
= ¦x
1
, . . . , x
l
¦.
q: A → ¸ Classiﬁcation rule.
f
y
: A →R Discriminant function associated with the class y ∈ ¸.
f: A →R Discriminant function of the binary classiﬁer.
[ [ Cardinality of a set.
  Euclidean norm.
det(()) Matrix determinant.
δ(i, j) Kronecker delta δ(i, j) = 1 for i = j and δ(i, j) = 0
otherwise.
_
Logical or.
7
Chapter 2
Linear Discriminant Function
The linear classiﬁcation rule q: A ⊆ R
n
→ ¸ = ¦1, 2, . . . , c¦ is composed of a set of
discriminant functions
f
y
(x) = ¸w
y
x) + b
y
, ∀y ∈ ¸ ,
which are linear with respect to both the input vector x ∈ R
n
and their parameter
vectors w ∈ R
n
. The scalars b
y
, ∀y ∈ ¸ introduce bias to the discriminant functions.
The input vector x ∈ R
n
is assigned to the class y ∈ ¸ its corresponding discriminant
function f
y
attains maximal value
y = argmax
y
∈Y
f
y
(x) = argmax
y
∈Y
(¸w
y
x) + b
y
) . (2.1)
In the particular binary case ¸ = ¦1, 2¦, the linear classiﬁer is represented by a
single discriminant function
f(x) = ¸w x) + b ,
given by the parameter vector w ∈ R
n
and bias b ∈ R. The input vector x ∈ R
n
is
assigned to class y ∈ ¦1, 2¦ as follows
q(x) =
_
1 if f(x) = ¸w x) + b ≥ 0 ,
2 if f(x) = ¸w x) + b < 0 .
(2.2)
The datatype used to describe the linear classiﬁer and the implementation of the
classiﬁer is described in Section 2.1. Following sections describe supervised learning
methods which determine the parameters of the linear classiﬁer based on available
knowledge. The implemented learning methods can be distinguished according to the
type of the training data:
I. The problem is described by a ﬁnite set T = ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ containing pairs
of observations x
i
∈ R
n
and corresponding class labels y
i
∈ ¸. The methods
using this type of training data is described in Section 2.2 and Section 2.3.
8
II. The parameters of class conditional distributions are (completely or partially)
known. The Anderson’s task and its generalization are representatives of such
methods (see Section 2.4).
The summary of implemented methods is given in Table 2.1.
References: The linear discriminant function is treated throughout for instance in
books [26, 6].
Table 2.1: Implemented methods: Linear discriminant function
linclass Linear classiﬁer (linear machine).
andrerr Classiﬁcation error of the Generalized Anderson’s
task.
androrig Original method to solve the AndersonBahadur’s
task.
eanders Epsilonsolution of the Generalized Anderson’s task.
ganders Solves the Generalized Anderson’s task.
ggradander Gradient method to solve the Generalized Anderson’s
task.
ekozinec Kozinec’s algorithm for epsoptimal separating hyper
plane.
mperceptr Perceptron algorithm to train multiclass linear ma
chine.
perceptron Perceptron algorithm to train binary linear classiﬁer.
fld Fisher Linear Discriminant.
fldqp Fisher Linear Discriminant using Quadratic Program
ming.
demo linclass Demo on the algorithms learning linear classiﬁers.
demo anderson Demo on Generalized Anderson’s task.
2.1 Linear classiﬁer
The STPRtool uses a speciﬁc structure array to describe the binary (see Table 2.2)
and the multiclass (see Table 2.3) linear classiﬁer. The implementation of the linear
classiﬁer itself provides function linclass.
Example: Linear classiﬁer
The example shows training of the Fisher Linear Discriminant which is the classical
example of the linear classiﬁer. The Riply’s data set riply trn.mat is used for training.
The resulting linear classiﬁer is then evaluated on the testing data riply tst.mat.
9
Table 2.2: Datatype used to describe binary linear classiﬁer.
Binary linear classiﬁer (structure array):
.W [n 1] The normal vector w ∈ R
n
of the separating hyper
plane f(x) = ¸w x) + b.
.b [1 1] Bias b ∈ R of the hyperplane.
.fun = ’linclass’ Identiﬁes function associated with this data type.
Table 2.3: Datatype used to describe multiclass linear classiﬁer.
Multiclass linear classiﬁer (structure array):
.W [n c] Matrix of parameter vectors w
y
∈ R
n
, y = 1, . . . , c of
the linear discriminant functions f
y
(x) = ¸w
y
x) +b.
.b [c 1] Parameters b
y
∈ R, y = 1, . . . , c.
.fun = ’linclass’ Identiﬁes function associated with this data type.
trn = load(’riply_trn’); % load training data
tst = load(’riply_tst’); % load testing data
model = fld( trn ); % train FLD classifier
ypred = linclass( tst.X, model ); % classify testing data
cerror( ypred, tst.y ) % evaluate testing error
ans =
0.1080
2.2 Linear separation of ﬁnite sets of vectors
The input training data T = ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ consists of pairs of observations
x
i
∈ R
n
and corresponding class labels y
i
∈ ¸ = ¦1, . . . , c¦. The implemented methods
solve the task of (i) training separating hyperplane for binary case c = 2, (ii) training
εoptimal hyperplane and (iii) training the multiclass linear classiﬁer. The deﬁnitions
of these tasks are given below.
The problem of training the binary (c = 2) linear classiﬁer (2.2) with zero training
error is equivalent to ﬁnding the hyperplane H = ¦x: ¸w x) +b = 0¦ which separates
the training vectors of the ﬁrst y = 1 and the second y = 2 class. The problem is
formally deﬁned as solving the set of linear inequalities
¸w x
i
) + b ≥ 0 , y
i
= 1 ,
¸w x
i
) + b < 0 , y
i
= 2 .
(2.3)
with respect to the vector w ∈ R
n
and the scalar b ∈ R. If the inequalities (2.3) have a
solution then the training data T is linearly separable. The Perceptron (Section 2.2.1)
10
and Kozinec’s (Section 2.2.2) algorithm are useful to train the binary linear classiﬁers
from the linearly separable data.
If the training data is linearly separable then the optimal separating hyperplane
can be deﬁned the solution of the following task
(w
∗
, b
∗
) = argmax
w,b
m(w, b)
= argmax
w,b
min
_
min
i∈I
1
¸w x
i
) + b
w
, min
i∈I
2
−
¸w x
i
) + b
w
_
,
(2.4)
where 1
1
= ¦i: y
i
= 1¦ and 1
2
= ¦i: y
i
= 2¦ are sets of indices. The optimal separating
hyperplane H
∗
= ¦x ∈ R
n
: ¸w
∗
x) + b
∗
= 0¦ separates training data T with maxi
mal margin m(w
∗
, b
∗
). This task is equivalent to training the linear Support Vector
Machines (see Section 5). The optimal separating hyperplane cannot be found exactly
except special cases. Therefore the numerical algorithms seeking the approximate so
lution are applied instead. The εoptimal solution is deﬁned as the vector w and scalar
b such that the inequality
m(w
∗
, b
∗
) −m(w, b) ≤ ε , (2.5)
holds. The parameter ε ≥ 0 deﬁnes closeness to the optimal solution in terms of the
margin. The εoptimal solution can be found by Kozinec’s algorithm (Section 2.2.2).
The problem of training the multiclass linear classiﬁer c > 2 with zero training
error is formally stated as the problem of solving the set of linear inequalities
¸w
y
i
x
i
) + b
y
i
> ¸w
y
x
i
) + b
y
, i = 1, . . . , l , y
i
,= y , (2.6)
with respect to the vectors w
y
∈ R
n
, y ∈ ¸ and scalars b
y
∈ R, y ∈ ¸. The task (2.6)
for linearly separable training data T can be solved by the modiﬁed Perceptron (Sec
tion 2.2.3) or Kozinec’s algorithm. An interactive demo on the algorithms separating
the ﬁnite sets of vectors by a hyperplane is implemented in demo linclass.
2.2.1 Perceptron algorithm
The input is a set T = ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of binary labeled y
i
∈ ¦1, 2¦ training
vectors x
i
∈ R
n
. The problem of training the separating hyperplane (2.3) can be
formally rewritten to a simpler form
¸v z
i
) > 0 , i = 1, . . . , l , (2.7)
where the vector v ∈ R
n+1
is constructed as
v = [w; b] , (2.8)
11
and transformed training data Z = ¦z
1
, . . . , z
l
¦ are deﬁned as
z
i
=
_
[x
i
; 1] if y
i
= 1 ,
−[x
i
; 1] if y
i
= 2 .
(2.9)
The problem of solving (2.7) with respect to the unknown vector v ∈ R
n+1
is equivalent
to the original task (2.3). The parameters (w, b) of the linear classiﬁer are obtained
from the found vector v by inverting the transformation (2.8).
The Perceptron algorithm is an iterative procedure which builds a series of vectors
v
(0)
, v
(1)
, . . . , v
(t)
until the set of inequalities (2.7) is satisﬁed. The initial vector v
(0)
can
be set arbitrarily (usually v = 0). The Novikoﬀ theorem ensures that the Perceptron
algorithm stops after ﬁnite number of iterations t if the training data are linearly
separable. Perceptron algorithm is implemented in function perceptron.
References: Analysis of the Perceptron algorithm including the Novikoﬀ’s proof of
convergence can be found in Chapter 5 of the book [26]. The Perceptron from the
neural network perspective is described in the book [3].
Example: Training binary linear classiﬁer with Perceptron
The example shows the application of the Perceptron algorithm to ﬁnd the binary
linear classiﬁer for synthetically generated 2dimensional training data. The generated
training data contain 50 labeled vectors which are linearly separable with margin 1.
The found classiﬁer is visualized as a separating hyperplane (line in this case) H =
¦x ∈ R
2
: ¸w x) + b = 0¦. See Figure 2.1.
data = genlsdata( 2, 50, 1); % generate training data
model = perceptron( data ); % call perceptron
figure;
ppatterns( data ); % plot training data
pline( model ); % plot separating hyperplane
2.2.2 Kozinec’s algorithm
The input is a set T = ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of binary labeled y
i
∈ ¦1, 2¦ training
vectors x
i
∈ R
n
. The Kozinec’s algorithm builds a series of vectors w
(0)
1
, w
(1)
1
, . . . , w
(t)
1
and w
(0)
2
, w
(1)
2
, . . . , w
(t)
2
which converge to the vector w
∗
1
and w
∗
2
respectively. The
vectors w
∗
1
and w
∗
2
are the solution of the following task
w
∗
1
, w
∗
2
= argmin
w
1
∈X
1
,w
2
∈X
2
w
1
−w
2
 ,
where A
1
stands for the convex hull of the training vectors of the ﬁrst class A
1
=
¦x
i
: y
i
= 1¦ and A
2
for the convex hull of the second class likewise. The vector w
∗
=
12
−55 −50 −45 −40 −35 −30
−55
−50
−45
−40
−35
−30
Figure 2.1: Linear classiﬁer trained by the Perceptron algorithm
w
∗
1
−w
∗
2
and the bias b
∗
=
1
2
(w
∗
2

2
−w
∗
1

2
) determine the optimal hyperplane (2.4)
separating the training data with the maximal margin.
The Kozinec’s algorithm is proven to converge to the vectors w
∗
1
and w
∗
2
in inﬁnite
number of iterations t = ∞. If the εoptimal optimality stopping condition is used
then the Kozinec’s algorithm converges in ﬁnite number of iterations. The Kozinec’s
algorithm can be also used to solve a simpler problem of ﬁnding the separating hyper
plane (2.3). Therefore the following two stopping conditions are implemented:
I. The separating hyperplane (2.3) is sought for (ε < 0). The Kozinec’s algorithm is
proven to converge in a ﬁnite number of iterations if the separating hyperplane
exists.
II. The εoptimal hyperplane (2.5) is sought for (ε ≥ 0). Notice that setting ε = 0
force the algorithm to seek the optimal hyperplane which is generally ensured to
be found in an inﬁnite number of iterations t = ∞.
The Kozinec’s algorithm which trains the binary linear classiﬁer is implemented in the
function ekozinec.
References: Analysis of the Kozinec’s algorithm including the proof of convergence
can be found in Chapter 5 of the book [26].
Example: Training εoptimal hyperplane by the Kozinec’s algorithm
The example shows application of the Kozinec’s algorithm to ﬁnd the (ε = 0.01)
optimal hyperplane for synthetically generated 2dimensional training data. The gen
erated training data contains 50 labeled vectors which are linearly separable with
13
margin 1. The found classiﬁer is visualized (Figure 2.2) as a separating hyperplane
H = ¦x ∈ R
2
: ¸w x) +b = 0¦ (straight line in this case). It could be veriﬁed that the
found hyperplane has margin model.margin which satisﬁes the εoptimality condition.
data = genlsdata(2,50,1); % generate data
% run ekozinec
model = ekozinec(data, struct(’eps’,0.01));
figure;
ppatterns(data); % plot data
pline(model); % plot found hyperplane
−10 −8 −6 −4 −2 0 2 4 6 8 10
−10
−8
−6
−4
−2
0
2
4
6
8
Figure 2.2: εoptimal hyperplane trained with the Kozinec’s algorithm.
2.2.3 Multiclass linear classiﬁer
The input training data T = ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ consists of pairs of observations
x
i
∈ R
n
and corresponding class labels y
i
∈ ¸ = ¦1, . . . , c¦. The task is to design
the linear classiﬁer (2.1) which classiﬁes the training data T without error. The prob
lem of training (2.1) is equivalent to solving the set of linear inequalities (2.6). The
inequalities (2.6) can be formally transformed to a simpler set
¸v z
y
i
) > 0 , i = 1, . . . , l , y = ¸ ¸ ¦y
i
¦ , (2.10)
where the vector v ∈ R
(n+1)c
contains the originally optimized parameters
v = [w
1
; b
1
; w
2
; b
2
; . . . ; w
c
; b
c
] .
14
The set of vectors z
y
i
∈ R
(n+1)c
, i = 1, . . . , l, y ∈ ¸ ¸ ¦y
i
¦ is created from the training
set T such that
z
y
i
(j) =
⎧
⎨
⎩
[x
i
; 1] , for j = y
i
,
−[x
i
; 1] , for j = y ,
0 , otherwise.
(2.11)
where z
y
i
(j) stands for the jth slot between coordinates (n+1)(j −1)+1 and (n+1)j.
The described transformation is known as the Kesler’s construction.
The transformed task (2.10) can be solved by the Perceptron algorithm. The simple
form of the Perceptron updating rule allows to implement the transformation implicitly
without mapping the training data T into (n + 1)cdimensional space. The described
method is implemented in function mperceptron.
References: The Kesler’s construction is described in books [26, 6].
Example: Training multiclass linear classiﬁer by the Perceptron
The example shows application of the Perceptron rule to train the multiclass linear
classiﬁer for the synthetically generated 2dimensional training data pentagon.mat. A
user’s own data can be generated interactively using function createdata(’finite’,10)
(number 10 stands for number of classes). The found classiﬁer is visualized as its sep
arating surface which is known to be piecewise linear and the class regions should be
convex (see Figure 2.3).
data = load(’pentagon’); % load training data
model = mperceptron(data); % run training algorithm
figure;
ppatterns(data); % plot training data
pboundary(model); % plot decision boundary
2.3 Fisher Linear Discriminant
The input is a set T = ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of binary labeled y
i
∈ ¦1, 2¦ training
vectors x
i
∈ R
n
. Let 1
y
= ¦i: y
i
= y¦, y ∈ ¦1, 2¦ be sets of indices of training
vectors belonging to the ﬁrst y = 1 and the second y = 2 class, respectively. The class
separability in a direction w ∈ R
n
is deﬁned as
F(w) =
¸w S
B
w)
¸w S
W
w)
, (2.12)
where S
B
is the betweenclass scatter matrix
S
B
= (µ
1
−µ
2
)(µ
1
−µ
2
)
T
, µ
y
=
1
[1
y
[
i∈Iy
x
i
, y ∈ ¦1, 2¦ ,
15
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Figure 2.3: Multiclass linear classiﬁer trained by the Perceptron algorithm.
and S
W
is the within class scatter matrix deﬁned as
S
W
= S
1
+S
2
, S
y
=
i∈Yy
(x
i
−µ
y
)(x
i
−µ
y
)
T
, y ∈ ¦1, 2¦ .
In the case of the Fisher Linear Discriminant (FLD), the parameter vector w of the
linear discriminant function f(x) = ¸w x) + b is determined to maximize the class
separability criterion (2.12)
w = argmax
w
F(w
) = argmax
w
¸w
S
B
w
)
¸w
S
W
w
)
. (2.13)
The bias b of the linear rule must be determined based on another principle. The
STPRtool implementation, for instance, computes such b that the equality
¸w µ
1
) + b = −(¸w µ
2
) + b) ,
holds. The STPRtool contains two implementations of FLD:
I. The classical solution of the problem (2.13) using the matrix inversion
w = S
W
−1
(µ
1
−µ
2
) .
This case is implemented in the function fld.
16
II. The problem (2.13) can be reformulated as the quadratic programming (QP) task
w = argmin
w
¸w
S
W
w
) ,
subject to
¸w
(µ
1
−µ
2
)) = 2 .
The QP solver quadprog of the Matlab Optimization Toolbox is used in the
toolbox implementation. This method can be useful when the matrix inversion
is hard to compute. The method is implemented in the function fldqp.
References: The Fisher Linear Discriminant is described in Chapter 3 of the book [6]
or in the book [3].
Example: Fisher Linear Discriminant
The binary linear classiﬁer based on the Fisher Linear Discriminant is trained on
the Riply’s data set riply trn.mat. The QP formulation fldqp of FLD is used.
However, the function fldqp can be replaced by the standard approach implemented
in the function fld which would yield the same solution. The found linear classiﬁer is
visualized (see Figure 2.4) and evaluated on the testing data riply tst.mat.
trn = load(’riply_trn’); % load training data
model = fldqp(trn); % compute FLD
figure;
ppatterns(trn); pline(model); % plot data and solution
tst = load(’riply_tst’); % load testing data
ypred = linclass(tst.X,model); % classify testing data
cerror(ypred,tst.y) % compute testing error
ans =
0.1080
2.4 Generalized Anderson’s task
The classiﬁed object is described by the vector of observations x ∈ A ⊆ R
n
and a
binary hidden state y ∈ ¦1, 2¦. The class conditional distributions p
XY
(x[y), y ∈ ¦1, 2¦
are known to be multivariate Gaussian distributions. The parameters (µ
1
, Σ
1
) and
(µ
2
, Σ
2
) of these class distributions are unknown. However, it is known that the
parameters (µ
1
, Σ
1
) belong to a certain ﬁnite set of parameters ¦(µ
i
, Σ
i
): i ∈ 1
1
¦.
Similarly (µ
2
, Σ
2
) belong to a ﬁnite set ¦(µ
i
, Σ
i
): i ∈ 1
2
¦. Let q: A ⊆ R
n
→ ¦1, 2¦
17
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 2.4: Linear classiﬁer based on the Fisher Linear Discriminant.
be a binary linear classiﬁer (2.2) with discriminant function f(x) = ¸w x) + b. The
probability of misclassiﬁcation is deﬁned as
Err(w, b) = max
i∈I
1
∪I
2
ε(w, b, µ
i
, Σ
i
) ,
where ε(w, b, µ
i
, Σ
i
) is probability that the Gaussian random vector x with mean
vector µ
i
and the covariance matrix Σ
i
satisﬁes q(x) = 1 for i ∈ 1
2
or q(x) = 2 for
i ∈ 1
1
. In other words, it is the probability that the vector x will be misclassiﬁed by
the linear rule q.
The Generalized Anderson’s task (GAT) is to ﬁnd the parameters (w
∗
, b
∗
) of the
linear classiﬁer
q(x) =
_
1 , for f(x) = ¸w
∗
x) + b
∗
≥ 0 ,
2 , for f(x) = ¸w
∗
x) + b
∗
< 0 ,
such that the error Err(w
∗
, b
∗
) is minimal
(w
∗
, b
∗
) = argmin
w,b
Err(w, b) = argmin
w,b
max
i∈I
1
∪I
2
ε(w, b, µ
i
, Σ
i
) . (2.14)
The original Anderson’s task is a special case of (2.14) when [1
1
[ = 1 and [1
2
[ = 1.
The probability ε(w, b, µ
i
, Σ
i
) is proportional to the reciprocal of the Mahalanobis
distance r
i
between the (µ
i
, Σ
i
) and the nearest vector of the separating hyperplane
H = ¦x ∈ R
n
: ¸w x) + b = 0¦, i.e.,
r
i
= min
x∈H
¸(µ
i
−x) (Σ
i
)
−1
(µ
i
−x)) =
¸w µ
i
) + b
_
¸w Σ
i
w)
.
18
The exact relation between the probability ε(w, b, µ
i
, Σ
i
) and the corresponding Ma
halanobis distance r
i
is given by the integral
ε(w, b, µ
i
, Σ
i
) =
_
∞
r
i
1
√
2π
e
−
1
2
t
2
dt . (2.15)
The optimization problem (2.14) can be equivalently rewritten as
(w
∗
, b
∗
) = argmax
w,b
F(w, b) = argmax
w,b
min
i∈I
1
∪I
2
¸w µ
i
) + b
_
¸w Σ
i
w)
.
which is more suitable for optimization. The objective function F(w, b) is proven to
be convex in the region where the probability of misclassiﬁcation Err(w, b) is less than
0.5. However, the objective function F(w, b) is not diﬀerentiable.
The STPRtool contains implementations of the algorithm solving the original An
derson’s task as well as implementations of three diﬀerent approaches to solve the
Generalized Anderson’s task which are described bellow. An interactive demo on the
algorithms solving the Generalized Anderson’s task is implemented in demo anderson.
References: The original Anderson’s task was published in [1]. A detailed description
of the Generalized Anderson’s task and all the methods implemented in the STPRtool
is given in book [26].
2.4.1 Original Anderson’s algorithm
The algorithm is implemented in the function androrig. This algorithm solves the
original Anderson’s task deﬁned for two Gaussians only [1
1
[ = [1
2
[ = 1. In this case,
the optimal solution vector w ∈ R
n
is obtained by solving the following problem
w = ((1 −λ)Σ
1
+ λΣ
2
)
−1
(µ
1
−µ
2
) ,
1 −λ
λ
=
¸
¸w Σ
2
w)
¸w Σ
1
w)
,
where the scalar 0 < λ < 1 is unknown. The problem is solved by an iterative algorithm
which stops while the condition
¸
¸
¸
¸
γ −
1 −λ
λ
¸
¸
¸
¸
≤ ε , γ =
¸
¸w Σ
2
w)
¸w Σ
1
w)
,
is satisﬁed. The parameter ε deﬁnes the closeness to the optimal solution.
Example: Solving the original Anderson’s task
19
The original Anderson’s task is solved for the Riply’s data set riply trn.mat. The
parameters (µ
1
, Σ
1
) and (µ
2
, Σ
2
) of the Gaussian distribution of the ﬁrst and the
second class are obtained by the MaximumLikelihood estimation. The found linear
classiﬁer is visualized and the Gaussian distributions (see Figure 2.5). The classiﬁer is
validated on the testing set riply tst.mat.
trn = load(’riply_trn’); % load training data
distrib = mlcgmm(data); % estimate Gaussians
model = androrig(distrib); % solve Anderson’s task
figure;
pandr( model, distrib ); % plot solution
tst = load(’riply_tst’); % load testing data
ypred = linclass( tst.X, model ); % classify testing data
cerror( ypred, tst.y ) % evaluate error
ans =
0.1150
−0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
gauss 1
0.929
gauss 2
1.37
Figure 2.5: Solution of the original Anderson’s task.
20
2.4.2 General algorithm framework
The algorithm is implemented in the function ganders. This algorithm repeats the
following steps:
I. An improving direction (∆w, ∆b) in the parameter space is computed such that
moving along this direction increases the objective function F. The quadprog
function of the Matlab optimization toolbox is used to solve this subtask.
II. The optimal movement k along the direction (∆w, ∆b). The optimal movement k
is determined by 1Dsearch based on the cutting interval algorithm according to
the Fibonacci series. The number of interval cuttings of is an optional parameter
of the algorithm.
III. The new parameters are set
w
(t+1)
: = w
(t)
+ k∆w , b
(t+1)
: = b
(t)
+ k∆b .
The algorithm iterates until the minimal change in the objective function is less than
the prescribed ε, i.e., the condition
F(w
(t+1)
, b
(t+1)
) −F(w
(t)
, b
(t)
) > ε ,
is violated.
Example: Solving the Generalized Anderson’s task
The Generalized Anderson’s task is solved for the data mars.mat. Both the ﬁrst
and second class are described by three 2Ddimensional Gaussian distributions. The
found linear classiﬁer as well as the input Gaussian distributions are visualized (see
Figure 2.6).
distrib = load(’mars’); % load input Gaussians
model = ganders(distrib); % solve the GAT
figure;
pandr(model,distrib); % plot the solution
2.4.3 εsolution using the Kozinec’s algorithm
This algorithm is implemented in the function eanders. The εsolution is deﬁned as a
linear classiﬁer with parameters (w, b) such that
F(w, b) < ε
21
−5 −4 −3 −2 −1 0 1 2 3 4 5
−4
−3
−2
−1
0
1
2
3
4
5
gauss 1
0.0461
gauss 2
0.0483
gauss 3
0.0503
gauss 4
0.0421
gauss 5
0.047
gauss 6
0.0503
Figure 2.6: Solution of the Generalized Anderson’s task.
is satisﬁed. The prescribed ε ∈ (0, 0.5) is the desired upper bound on the probability
of misclassiﬁcation. The problem is transformed to the task of separation of two sets
of ellipsoids which is solved the Kozinec’s algorithm. The algorithm converges in ﬁnite
number of iterations if the given εsolution exists. Otherwise it can get stuck in an
endless loop.
Example: εsolution of the Generalized Anderson’s task
The εsolution of the generalized Anderson’s task is solved for the data mars.mat.
The desired maximal possible probability of misclassiﬁcation is set to 0.06 (it is 6%).
Both the ﬁrst and second class are described by three 2Ddimensional Gaussian distri
butions. The found linear classiﬁer and the input Gaussian distributions are visualized
(see Figure 2.7).
distrib = load(’mars’); % load Gaussians
options = struct(’err’,0.06’); % maximal desired error
model = eanders(distrib,options); % training
figure; pandr(model,distrib); % visualization
2.4.4 Generalized gradient optimization
This algorithm is implemented in the function ggradandr. The objective function
F(w, b) is convex but not diﬀerentiable thus the standard gradient optimization cannot
22
−5 −4 −3 −2 −1 0 1 2 3 4 5
−4
−3
−2
−1
0
1
2
3
4
5
gauss 1
0.0473
gauss 2
0.0496
gauss 3
0.0516
gauss 4
0.0432
gauss 5
0.0483
gauss 6
0.0516
Figure 2.7: εsolution of the Generalized Anderson’s task with maximal 0.06 probability
of misclassiﬁcation.
be used. However, the generalized gradient optimization method can be applied. The
algorithm repeats the following two steps:
I. The generalized gradient (∆w, ∆b) is computed.
II. The optimized parameters are updated
w
(t+1)
: = w
(t)
+ ∆w , b
(t+1)
= b
(t)
+ ∆b .
The algorithm iterates until the minimal change in the objective function is less than
the prescribed ε, i.e., the condition
F(w
(t+1)
, b
(t+1)
) −F(w
(t)
, b
(t)
) > ε ,
is violated. The maximal number of iterations can be used as the stopping condition
as well.
Example: Solving the Generalized Anderson’s task using the generalized
gradient optimization
The generalized Anderson’s task is solved for the data mars.mat. Both the ﬁrst
and second class are described by three 2Ddimensional Gaussian distributions. The
maximal number of iterations is limited to tmax=10000. The found linear classiﬁer as
well as the input Gaussian distributions are visualized (see Figure 2.8).
23
distrib = load(’mars’); % load Gaussians
options = struct(’tmax’,10000); % max # of iterations
model = ggradandr(distrib,options); % training
figure;
pandr( model, distrib ); % visualization
−5 −4 −3 −2 −1 0 1 2 3 4 5
−4
−3
−2
−1
0
1
2
3
4
5
gauss 1
0.0475
gauss 2
0.0498
gauss 3
0.0518
gauss 4
0.0434
gauss 5
0.0485
gauss 6
0.0518
Figure 2.8: Solution of the Generalized Anderson’s task after 10000 iterations of the
generalized gradient optimization algorithm.
24
Chapter 3
Feature extraction
The feature extraction step consists of mapping the input vector of observations x ∈ R
n
onto a new feature description z ∈ R
m
which is more suitable for given task. The linear
projection
z = W
T
x +b , (3.1)
is commonly used. The projection is given by the matrix W [n m] and bias vector
b ∈ R
m
. The Principal Component Analysis (PCA) (Section 3.1) is a representative of
the unsupervised learning method which yields the linear projection (3.1). The Linear
Discriminant Analysis (LDA) (Section 3.2) is a representative of the supervised learning
method yielding the linear projection. The linear data projection is implemented in
function linproj (See examples in Section 3.1 and Section 3.2). The data type used
to describe the linear data projection is deﬁned in Table 3.
The quadratic data mapping projects the input vector x ∈ R
n
onto the feature
space R
m
where m = n(n + 3)/2. The quadratic mapping φ: R
n
→R
m
is deﬁned as
φ(x) = [ x
1
, x
2
, . . . , x
n
,
x
1
x
1
, x
1
x
2
, . . . , x
1
x
n
,
x
2
x
2
, . . . , x
2
x
n
,
.
.
.
x
n
x
n
]
T
,
(3.2)
where x
i
, i = 1, . . . , n are entries of the vector x ∈ R
n
. The quadratic data mapping
is implemented in function qmap.
The kernel functions allow for nonlinear extensions of the linear feature extrac
tion methods. In this case, the input vectors x ∈ R
n
are mapped into a new higher
dimensional feature space T in which the linear methods are applied. This yields a
nonlinear (kernel) projection of data which has generally deﬁned as
z = A
T
k(x) +b , (3.3)
25
where A [l m] is a parameter matrix and b ∈ R
m
a parameter vector. The vector
k(x) = [k(x, x
1
), . . . , k(x, x
l
)]
T
is a vector of kernel functions centered in the training
data or their subset. The x ∈ R
n
stands for the input vector and z ∈ R
m
is the
vector of extracted features. The extracted features are projections of φ(x) onto m
vectors (or functions) from the feature space T. The Kernel PCA (Section 3.3) and
the Generalized Discriminant Analysis (GDA) (Section 3.5) are representatives of the
feature extraction methods yielding the kernel data projection (3.3). The kernel data
projection is implemented in the function kernelproj (See examples in Section 3.3
and Section 3.5). The data type used to describe the kernel data projection is deﬁned
in Table 3. The summary of implemented methods for feature extraction is given in
Table 3.1
Table 3.1: Implemented methods for feature extraction
linproj Linear data projection.
kerproj Kernel data projection.
qmap Quadratic data mapping.
lin2quad Merges linear rule and quadratic mapping.
lin2svm Merges linear rule and kernel projection.
lda Linear Discriminant Analysis.
pca Principal Component Analysis.
pcarec Computes reconstructed vector after PCA projection.
gda Generalized Discriminant Analysis.
greedykpca Greedy Kernel Principal Component Analysis.
kpca Kernel Principal Component Analysis.
kpcarec Reconstructs image after kernel PCA.
demo pcacomp Demo on image compression using PCA.
Table 3.2: Datatype used to describe linear data projection.
Linear data projection (structure array):
.W [n m] Projection matrix W = [w
1
, . . . , w
m
].
.b [m1] Bias vector b.
.fun = ’linproj’ Identiﬁes function associated with this data type.
3.1 Principal Component Analysis
Let T
X
= ¦x
1
, . . . , x
l
¦ be a set of training vectors from the ndimensional input space
R
n
. The set of vectors T
Z
= ¦z
1
, . . . , z
l
¦ is a lower dimensional representation of the
26
Table 3.3: Datatype used to describe kernel data projection.
Kernel data projection (structure array):
.Alpha [l m] Parameter matrix A [l m].
.b [m1] Bias vector b.
.sv.X [n l] Training vectors ¦x
1
, . . . , x
l
¦.
.options.ker [string] Kernel identiﬁer.
.options.arg [1 p] Kernel argument(s).
.fun = ’kernelproj’ Identiﬁes function associated with this data type.
input training vectors T
X
in the mdimensional space R
m
. The vectors T
Z
are obtained
by the linear orthonormal projection
z = W
T
x +b , (3.4)
where the matrix W [n m] and the vector b [m 1] are parameters of the projec
tion. The reconstructed vectors T
˜
X
= ¦ ˜ x
1
, . . . , ˜ x
l
¦ are computed by the linear back
projection
˜ x = W(z −b) , (3.5)
obtained by inverting (3.4). The mean square reconstruction error
ε
MS
(W, b) =
1
l
l
i=1
x
i
− ˜ x
i

2
, (3.6)
is a function of the parameters of the linear projections (3.4) and (3.5). The Principal
Component Analysis (PCA) is the linear orthonormal projection (3.4) which allows
for the minimal mean square reconstruction error (3.6) of the training data T
X
. The
parameters (W, b) of the linear projection are the solution of the optimization task
(W, b) = argmin
W
,b
ε
MS
(W
, b
) , (3.7)
subject to
¸w
i
w
j
) = δ(i, j) , ∀i, j ,
where w
i
, i = 1, . . . , m are column vectors of the matrix W = [w
1
, . . . , w
m
] and
δ(i, j) is the Kronecker delta function. The solution of the task (3.7) is the matrix
W = [w
1
, . . . , w
m
] containing the m eigenvectors of the sample covariance matrix
which have the largest eigen values. The vector b equals to W
T
µ, where µ is the
sample mean of the training data. The PCA is implemented in the function pca. A
demo showing the use of PCA for image compression is implemented in the script
demo pcacomp.
27
References: The PCA is treated throughout for instance in books [13, 3, 6].
Example: Principal Component Analysis
The input data are synthetically generated from the 2dimensional Gaussian dis
tribution. The PCA is applied to train the 1dimensional subspace to approximate
the input data with the minimal reconstruction error. The training data, the recon
structed data and the 1dimensional subspace given by the ﬁrst principal component
are visualized (see Figure 3.1).
X = gsamp([5;5],[1 0.8;0.8 1],100); % generate data
model = pca(X,1); % train PCA
Z = linproj(X,model); % lower dim. proj.
XR = pcarec(X,model); % reconstr. data
figure; hold on; axis equal; % visualization
h1 = ppatterns(X,’kx’);
h2 = ppatterns(XR,’bo’);
[dummy,mn] = min(Z);
[dummy,mx] = max(Z);
h3 = plot([XR(1,mn) XR(1,mx)],[XR(2,mn) XR(2,mx)],’r’);
legend([h1 h2 h3], ...
’Input vectors’,’Reconstructed’, ’PCA subspace’);
2 3 4 5 6 7 8
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5
Input vectors
Reconstructed
PCA subspace
Figure 3.1: Example on the Principal Component Analysis.
28
3.2 Linear Discriminant Analysis
Let T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦, x
i
∈ R
n
, y ∈ ¸ = ¦1, 2, . . . , c¦ be labeled set
of training vectors. The withinclass S
W
and betweenclass S
B
scatter matrices are
described as
S
W
=
y∈Y
S
y
, S
y
=
i∈Iy
(x
i
−µ
i
)(x
i
−µ
i
)
T
,
S
B
=
y∈Y
[1
y
[(µ
y
−µ)(µ
y
−µ)
T
,
(3.8)
where the total mean vector µ and the class mean vectors µ
y
, y ∈ ¸ are deﬁned as
µ =
1
l
l
i=1
x
i
, µ
y
=
1
[1
y
[
i∈Iy
x
i
, y ∈ ¸ .
The goal of the Linear Discriminant Analysis (LDA) is to train the linear data projec
tion
z = W
T
x ,
such that the class separability criterion
F(W) =
det(
˜
S
B
)
det(
˜
S
W
)
=
det(S
B
)
det(S
W
)
,
is maximized. The
˜
S
B
is the betweenclass and
˜
S
W
is the withinclass scatter matrix
of projected data which is deﬁned similarly to (3.8).
References: The LDA is treated for instance in books [3, 6].
Example 1: Linear Discriminant Analysis
The LDA is applied to extract 2 features from the Iris data set iris.mat which
consists of the labeled 4dimensional data. The data after the feature extraction step
are visualized in Figure 3.2.
orig_data = load(’iris’); % load input data
model = lda(orig_data,2); % train LDA
ext_data = linproj(orig_data,model); % feature extraction
figure; ppatterns(ext_data); % plot ext. data
Example 2: PCA versus LDA
This example shows why the LDA is superior to the PCA when features good for
classiﬁcation are to be extracted. The synthetical data are generated from the Gaussian
mixture model. The LDA and PCA are trained on the generated data. The LDA and
29
−1.5 −1 −0.5 0 0.5 1 1.5
−0.5
−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
0.4
0.5
Figure 3.2: Example shows 2dimensional data extracted from the originally 4
dimensional Iris data set using Linear Discriminant Analysis.
PCA directions are displayed as connection of the vectors projected to the trained LDA
and PCA subspaces and reprojected back to the input space (see Figure 3.3a). The
extracted data using the LDA and PCA model are displayed with the Gaussians ﬁtted
by the MaximumLikelihood (see Figure 3.3b).
% Generate data
distrib.Mean = [[5;4] [4;5]]; % mean vectors
distrib.Cov(:,:,1) = [1 0.9; 0.9 1]; % 1st covariance
distrib.Cov(:,:,2) = [1 0.9; 0.9 1]; % 2nd covariance
distrib.Prior = [0.5 0.5]; % Gaussian weights
data = gmmsamp(distrib,250); % sample data
lda_model = lda(data,1); % train LDA
lda_rec = pcarec(data.X,lda_model);
lda_data = linproj(data,lda_model);
pca_model = pca(data.X,1); % train PCA
pca_rec = pcarec(data.X,pca_model);
pca_data = linproj( data,pca_model);
figure; hold on; axis equal; % visualization
ppatterns(data);
30
h1 = plot(lda_rec(1,:),lda_rec(2,:),’r’);
h2 = plot(pca_rec(1,:),pca_rec(2,:),’b’);
legend([h1 h2],’LDA direction’,’PCA direction’);
figure; hold on;
subplot(2,1,1); title(’LDA’); ppatterns(lda_data);
pgauss(mlcgmm(lda_data));
subplot(2,1,2); title(’PCA’); ppatterns(pca_data);
pgauss(mlcgmm(pca_data));
3.3 Kernel Principal Component Analysis
The Kernel Principal Component Analysis (Kernel PCA) is the nonlinear extension of
the ordinary linear PCA. The input training vectors T
X
= ¦x
1
, . . . , x
l
¦, x
i
∈ A ⊆ R
n
are mapped by φ: A → T to a high dimensional feature space T. The linear PCA
is applied on the mapped data T
Φ
= ¦φ(x
1
), . . . , φ(x
l
)¦. The computation of the
principal components and the projection on these components can be expressed in
terms of dot products thus the kernel functions k: A A →R can be employed. The
kernel PCA trains the kernel data projection
z = A
T
k(x) +b , (3.9)
such that the reconstruction error
ε
KMS
(A, b) =
1
l
l
i=1
φ(x
i
) −
˜
φ(x
i
)
2
, (3.10)
is minimized. The reconstructed vector
˜
φ(x) is given as a linear combination of the
mapped data T
Φ
˜
φ(x) =
l
i=1
β
i
φ(x
i
) , β = A(z −b) . (3.11)
In contrast to the linear PCA, the explicit projection from the feature space T to the
input space A usually do not exist. The problem is to ﬁnd the vector x ∈ A its image
φ(x) ∈ T well approximates the reconstructed vector
˜
φ(x) ∈ T. This procedure is
implemented in the function kpcarec for the case of the Radial Basis Function (RBF)
kernel. The procedure consists of the following step:
I. Project input vector x
in
∈ R
n
onto its lower dimensional representation z ∈ R
m
using (3.9).
31
II. Computes vector x
out
∈ R
n
which is good preimage of the reconstructed vector
˜
φ(x
in
), such that
x
out
= argmin
x
φ(x) −
˜
φ(x
in
)
2
.
References: The kernel PCA is described at length in book [29, 30].
Example: Kernel Principal Component Analysis
The example shows using the kernel PCA for data denoising. The input data are
synthetically generated 2dimensional vectors which lie on a circle and are corrupted by
the Gaussian noise. The kernel PCA model with RBF kernel is trained. The function
kpcarec is used to ﬁnd preimages of the reconstructed data (3.11). The result is
visualized in Figure 3.4.
X = gencircledata([1;1],5,250,1); % generate circle data
options.ker = ’rbf’; % use RBF kernel
options.arg = 4; % kernel argument
options.new_dim = 2; % output dimension
model = kpca(X,options); % compute kernel PCA
XR = kpcarec(X,model); % compute reconstruced data
figure; % Visualization
h1 = ppatterns(X);
h2 = ppatterns(XR, ’+r’);
legend([h1 h2],’Input vectors’,’Reconstructed’);
3.4 Greedy kernel PCA algorithm
The Greedy Kernel Principal Analysis (Greedy kernel PCA) is an eﬃcient algorithm
to compute the ordinary kernel PCA. Let T
X
= ¦x
1
, . . . , x
l
¦, x
i
∈ R
n
be the set of
input training vectors. The goal is to train the kernel data projection
z = A
T
k
S
(x) +b , (3.12)
where A [d m] is the parameter matrix, b [m 1] is the parameter vector and
k
S
= [k(x, s
1
), . . . , k(x, s
l
)]
T
are the kernel functions centered in the vectors T
S
=
¦s
1
, . . . , s
d
¦. The vector set T
S
is a subset of training data T
X
. In contrast to the
ordinary kernel PCA, the subset T
S
does not contain all the training vectors T
X
thus
the complexity of the projection (3.12) is reduced compared to (3.9). The objective of
the Greedy kernel PCA is to minimize the reconstruction error (3.11) while the size d
of the subset T
S
is kept small. The Greedy kernel PCA algorithm is implemented in
the function greedykpca.
32
References: The implemented Greedy kernel PCA algorithm is described in [11].
Example: Greedy Kernel Principal Component Analysis
The example shows using kernel PCA for data denoising. The input data are
synthetically generated 2dimensional vectors which lie on a circle and are corrupted
with the Gaussian noise. The Greedy kernel PCA model with RBF kernel is trained.
The number of vectors deﬁning the kernel projection (3.12) is limited to 25. In contrast
the ordinary kernel PCA (see experiment in Section 3.3) requires all the 250 training
vectors. The function kpcarec is used to ﬁnd preimages of the reconstructed data
obtained by the kernel PCA projection. The training and reconstructed vectors are
visualized in Figure 3.5. The vector of the selected subset T
X
are visualized as well.
X = gencircledata([1;1],5,250,1); % generate training data
options.ker = ’rbf’; % use RBF kernel
options.arg = 4; % kernel argument
options.new_dim = 2; % output dimension
options.m = 25; % size of sel. subset
model = greedykpca(X,options); % run greedy algorithm
XR = kpcarec(X,model); % reconstructed vectors
figure; % visualization
h1 = ppatterns(X);
h2 = ppatterns(XR,’+r’);
h3 = ppatterns(model.sv.X,’ob’,12);
legend([h1 h2 h3], ...
’Training set’,’Reconstructed set’,’Selected subset’);
3.5 Generalized Discriminant Analysis
The Generalized Discriminant Analysis (GDA) is the nonlinear extension of the ordi
nary Linear Discriminant Analysis (LDA). The input training data T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦,
x
i
∈ A ⊆ R
n
, y ∈ ¸ = ¦1, . . . , c¦ are mapped by φ: A → T to a high dimen
sional feature space T. The ordinary LDA is applied on the mapped data T
ΦY
=
¦(φ(x
1
), y
1
), . . . , (φ(x
l
), y
l
)¦. The computation of the projection vectors and the pro
jection on these vectors can be expressed in terms of dot products thus the kernel
functions k: A A → R can be employed. The resulting kernel data projection is
deﬁned as
z = A
T
k(x) +b ,
33
where A [l m] is the matrix and b [m 1] the vector trained by the GDA. The
parameters (A, b) are trained to increase the betweenclass scatter and decrease the
withinclass scatter of the extracted data T
ZY
= ¦(z
1
, y
1
), . . . , (z
l
, y
l
)¦, z
i
∈ R
m
. The
GDA is implemented in the function gda.
References: The GDA was published in the paper [2].
Example: Generalized Discriminant Analysis
The GDA is trained on the Iris data set iris.mat which contains 3 classes of
4dimensional data. The RBF kernel with kernel width σ = 1 is used. Notice, that
setting σ too low leads to the perfect separability of the training data T
XY
but the kernel
projection is heavily overﬁtted. The extracted data ¸
ZY
are visualized in Figure 3.6.
orig_data = load(’iris’); % load data
options.ker = ’rbf’; % use RBF kernel
options.arg = 1; % kernel arg.
options.new_dim = 2; % output dimension
model = gda(orig_data,options); % train GDA
ext_data = kernelproj(orig_data,model); % feature ext.
figure; ppatterns(ext_data); % visualization
3.6 Combining feature extraction and linear classi
ﬁer
Let T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ be a training set of observations x
i
∈ A ⊆ R
n
and
corresponding hidden states y
i
∈ ¸ = ¦1, . . . , c¦. Let φ: A →R
m
be a mapping
φ(x) = [φ
1
(x), . . . , φ
m
(x)]
T
deﬁning a nonlinear feature extraction step (e.g., quadratic data mapping or kernel
projection). The feature extraction step applied to the input data T
XY
yields extracted
data T
ΦY
= ¦(φ(x
1
), y
1
), . . . , (φ(x
l
), y
l
)¦. Let f(x) = ¸w φ(x)) + b be a linear
discriminant function found by an arbitrary training method on the extracted data
T
ΦY
. The function f can be seen as a nonlinear function of the input vectors x ∈ A,
as
f(x) = ¸w φ(x)) + b =
m
i=1
w
i
φ
i
(x) + b .
In the case of the quadratic data mapping (3.2) used as the feature extraction step,
the discriminant function can be written in the following form
f(x) = ¸x Ax) +¸x b) + c . (3.13)
34
The discriminant function (3.13) deﬁnes the quadratic classiﬁer described in Section 6.4.
Therefore the quadratic classiﬁer can be trained by (i) applying the quadratic data
mapping qmap, (ii) training a linear rule on the mapped data (e.g., perceptron, fld)
and (iii) combining the quadratic mapping and the linear rule. The combining is
implement in function lin2quad.
In the case of the kernel projection (3.3) applied as the feature extraction step, the
resulting discriminant function has the following form
f(x) = ¸α k(x)) + b . (3.14)
The discriminant function (3.14) deﬁnes the kernel (SVM type) classiﬁer described
in Section 5. This classiﬁer can be trained by (i) applying the kernel projection
kernelproj, (ii) training a linear rule on the extracted data and (iii) combing the kernel
projection and the linear rule. The combining is implemented in function lin2svm.
Example: Quadratic classiﬁer trained the Perceptron
This example shows how to train the quadratic classiﬁer using the Perceptron al
gorithm. The Perceptron algorithm produces the linear rule only. The input training
data are linearly nonseparable but become separable after the quadratic data map
ping is applied. Function lin2quad is used to compute the parameters of the quadratic
classiﬁer explicitly based on the found linear rule in the feature space and the mapping.
Figure 3.7 shows the decision boundary of the quadratic classiﬁer.
% load training data living in the input space
input_data = load(’vltava’);
% map data to the feature space
map_data = qmap(input_data);
% train linear rule in the feature sapce
lin_model = perceptron(map_data);
% compute parameters of the quadratic classifier
quad_model = lin2quad(lin_model);
% visualize the quadratic classifier
figure; ppatterns(input_data);
pboundary(quad_model);
Example: Kernel classiﬁer from the linear rule
This example shows how to train the kernel classiﬁer using arbitrary training algo
rithm for linear rule. The Fisher Linear Discriminant was used in this example. The
35
greedy kernel PCA algorithm was applied as the feature extraction step. The Riply’s
data set was used for training and evaluation. Figure 3.8 shows the found kernel clas
siﬁer, the training data and vectors used in the kernel expansion of the discriminant
function.
% load input data
trn = load(’riply_trn’);
% train kernel PCA by greedy algorithm
options = struct(’ker’,’rbf’,’arg’,1,’new_dim’,10);
kpca_model = greedykpca(trn.X,options);
% project data
map_trn = kernelproj(trn,kpca_model);
% train linear classifier
lin_model = fld(map_trn);
% combine linear rule and kernel PCA
kfd_model = lin2svm(kpca_model,lin_model);
% visualization
figure;
ppatterns(trn); pboundary(kfd_model);
ppatterns(kfd_model.sv.X,’ok’,13);
% evaluation on testing data
tst = load(’riply_tst’);
ypred = svmclass(tst.X,kfd_model);
cerror(ypred,tst.y)
ans =
0.0970
36
0 1 2 3 4 5 6 7 8
1
2
3
4
5
6
7
LDA direction
PCA direction
(a)
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
LDA
−4 −3 −2 −1 0 1 2 3 4 5
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
PCA
(b)
Figure 3.3: Comparison between PCA and LDA. Figure (a) shows input data and
found LDA (red) and PCA (blue) directions onto which the data are projected. Figure
(b) shows the class conditional Gaussians estimated from data projected data onto
LDA and PCA directions.
37
−8 −6 −4 −2 0 2 4 6 8 10
−8
−6
−4
−2
0
2
4
6
8
10
Input vectors
Reconstructed
Figure 3.4: Example on the Kernel Principal Component Analysis.
−8 −6 −4 −2 0 2 4 6 8 10
−6
−4
−2
0
2
4
6
8
Training set
Reconstructed set
Selected set
Figure 3.5: Example on the Greedy Kernel Principal Component Analysis.
38
−0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3
−0.2
−0.15
−0.1
−0.05
0
0.05
0.1
0.15
Figure 3.6: Example shows 2dimensional data extracted from the originally 4
dimensional Iris data set using the Generalized Discriminant Analysis. In contrast
to LDA (see Figure 3.2 for comparison), the class clusters are more compact but at the
expense of a higher risk of overﬁtting.
−0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
Figure 3.7: Example shows quadratic classiﬁer found by the Perceptron algorithm on
the data mapped to the feature by the quadratic mapping.
39
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 3.8: Example shows the kernel classiﬁer found by the Fisher Linear Discriminant
algorithm on the data mapped to the feature by the kernel PCA. The greedy kernel
PCA was used therefore only a subset of training data was used to determine the kernel
projection.
40
Chapter 4
Density estimation and clustering
The STPRtool represents a probability distribution function P
X
(x), x ∈ R
n
as a
structure array which must contain ﬁeld .fun (see Table 4.2). The ﬁeld .fun speciﬁes
function which is used to evaluate P
X
(x) for given set of realization ¦x
1
, . . . , x
l
¦ of a
random vector. Let the structure model represent a given distribution function. The
Matlab command
y = feval(X,model.fun)
is used to evaluate y
i
= P
X
(x
i
). The matrix X [n l] contains column vectors
¦x
1
, . . . , x
l
¦. The vector y [1 l] contains values of the distribution function. In
particular, the Gaussian distribution and the Gaussian mixture models are described
in Section 4.1. The summary of implemented methods for dealing with probability
density estimation and clustering is given in Table 4.1
4.1 Gaussian distribution and Gaussian mixture model
The value of the multivariate Gaussian probability distribution function in the vector
x ∈ R
n
is given by
P
X
(x) =
1
(2π)
n
2
_
det(Σ)
exp(−
1
2
(x −µ)
T
Σ
−1
(x −µ)) , (4.1)
where µ [n 1] is the mean vector and Σ [n n] is the covariance matrix. The
STPRtool uses speciﬁc datatype to describe a set of Gaussians which can be labeled,
i.e., each pair (µ
i
, Σ
i
) is assigned to the class y
i
∈ ¸. The covariance matrix Σ is
always represented as the square matrix [n n], however, the shape of the matrix can
be indicated in the optimal ﬁeld .cov type (see Table 4.3). The datatype is described
in Table 4.4. The evaluation of the Gaussian distribution is implemented in the function
41
Table 4.1: Implemented methods: Density estimation and clustering
gmmsamp Generates sample from Gaussian mixture model.
gsamp Generates sample from Gaussian distribution.
kmeans Kmeans clustering algorithm.
pdfgauss Evaluates for multivariate Gaussian distribution.
pdfgauss Evaluates Gaussian mixture model.
sigmoid Evaluates sigmoid function.
emgmm ExpectationMaximization Algorithm for Gaussian
mixture model.
melgmm Maximizes Expectation of LogLikelihood for Gaus
sian mixture.
mlcgmm Maximal Likelihood estimation of Gaussian mixture
model.
mlsigmoid Fitting a sigmoid function using ML estimation.
mmgauss Minimax estimation of Gaussian distribution.
rsde Reduced Set Density Estimator.
demo emgmm Demo on ExpectationMaximization (EM) algorithm.
demo mmgauss Demo on minimax estimation for Gaussian.
demo svmpout Fitting a posteriori probability to SVM output.
Table 4.2: Datatype used to represent probability distribution function.
Probability Distribution Function (structure array):
.fun [string] The name of a function which evaluates probability
distribution y = feval( X, model.fun ).
pdfgauss. The random sampling from the Gaussian distribution is implemented in the
function gsamp.
The Gaussian Mixture Model (GMM) is weighted sum of the Gaussian distributions
P
X
(x) =
y∈Y
P
Y
(y)P
XY
(x[y) , (4.2)
where P
Y
(y), y ∈ ¸ = ¦1, . . . , c¦ are weights and P
XY
(x[y) is the Gaussian distribu
tion (4.1) given by parameters (µ
y
, Σ
y
). The datatype used to represent the GMM
is described in Table 4.5. The evaluation of the GMM is implemented in the function
pdfgmm. The random sampling from the GMM is implemented in the function gmmsamp.
Example: Sampling from the Gaussian distribution
The example shows how to represent the Gaussian distribution and how to sample
data it. The sampling from univariate and bivariate Gaussians is shown. The distribu
tion function of the univariate Gaussian is visualized and compared to the histogram
42
Table 4.3: Shape of covariance matrix.
Parameter cov type:
’full’ Full covariance matrix.
’diag’ Diagonal covariance matrix.
’spherical’ Spherical covariance matrix.
Table 4.4: Datatype used to represent the labeled set of Gaussians.
Labeled set of Gaussians (structure array):
.Mean [n k] Mean vectors ¦µ
1
, . . . , µ
k
¦.
.Cov [n n k] Covariance matrices ¦Σ
1
, . . . , Σ
k
¦.
.y [1 k] Labels ¦y
1
, . . . , y
k
¦ assigned to Gaussians.
.cov type [string] Optional ﬁeld which indicates shape of the covariance
matrix (see Table 4.3).
.fun = ’pdfgauss’ Identiﬁes function associated with this data type.
of the sample data (see Figure 4.1a). The isoline of the distribution function of the
bivariate Gaussian is visualized together with the sample data (see Figure 4.1b).
% univariate case
model = struct(’Mean’,1,’Cov’,2); % Gaussian parameters
figure; hold on;
% plot pdf. of the Gaussisan
plot([4:0.1:5],pdfgauss([4:0.1:5],model),’r’);
[Y,X] = hist(gsamp(model,500),10); % compute histogram
bar(X,Y/500); % plot histogram
% bivariate case
model.Mean = [1;1]; % Mean vector
model.Cov = [1 0.6; 0.6 1]; % Covariance matrix
figure; hold on;
ppatterns(gsamp(model,250)); % plot sampled data
pgauss(model); % plot shape
4.2 MaximumLikelihood estimation of GMM
Let the probability distribution
P
XY Θ
(x, y[θ) = P
XY Θ
(x[y, θ)P
Y Θ
(y[θ) ,
43
Table 4.5: Datatype used to represent the Gaussian Mixture Model.
Gaussian Mixture Model (structure array):
.Mean [n c] Mean vectors ¦µ
1
, . . . , µ
c
¦.
.Cov [n n c] Covariance matrices ¦Σ
1
, . . . , Σ
c
¦.
.Prior [c 1] Weights of Gaussians ¦P
K
(1), . . . , P
K
(c)¦.
.fun = ’pdfgmm’ Identiﬁes function associated with this data type.
−4 −3 −2 −1 0 1 2 3 4 5 6
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
−2 −1 0 1 2 3 4 5
−3
−2
−1
0
1
2
3
4
5
gauss 1
0.127
(a) (b)
Figure 4.1: Sampling from the univariate (a) and bivariate (b) Gaussian distribution.
be known up to parameters θ ∈ Θ. The vector of observations x ∈ R
n
and the hidden
state y ∈ ¸ = ¦1, . . . , c¦ are assumed to be realizations of random variables which
are independent and identically distributed according to the P
XY Θ
(x, y[θ). The con
ditional probability distribution P
XY Θ
(x[y, θ) is assumed to be Gaussian distribution.
The θ involves parameters (µ
y
, Σ
y
), y ∈ ¸ of Gaussian components and the values
of the discrete distribution P
Y Θ
(y[θ), y ∈ ¸. The marginal probability P
XΘ
(x[θ) is
the Gaussian mixture model (4.2). The MaximumLikelihood estimation of the pa
rameters θ of the GMM for the case of complete and incomplete data is described in
Section 4.2.1 and Section 4.2.2, respectively.
4.2.1 Complete data
The input is a set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ which contains both vectors of ob
servations x
i
∈ R
n
and corresponding hidden states y
i
∈ ¸. The pairs (x
i
, y
i
) are
assumed to samples from random variables which are independent and identically dis
tributed (i.i.d.) according to P
XY Θ
(x, y[θ). The logarithm of probability P(T
XY
[θ)
is referred to as the loglikelihood L(θ[T
XY
) of θ with respect to T
XY
. Thanks to the
44
i.i.d. assumption the loglikelihood can be factorized as
L(θ[T
XY
) = log P(T
XY
[θ) =
l
i=1
log P
XY Θ
(x
i
, y
i
[θ) .
The problem of MaximumLikelihood (ML) estimation from the complete data T
XY
is
deﬁned as
θ
∗
= argmax
θ∈Θ
L(θ[T
XY
) = argmax
θ∈Θ
l
i=1
log P
XY Θ
(x
i
, y
i
[θ) . (4.3)
The ML estimate of the parameters θ of the GMM from the complete data set T
XY
has an analytical solution. The computation of the ML estimate is implemented in the
function mlcgmm.
Example: MaximumLikelihood estimate from complete data
The parameters of the Gaussian mixture model is estimated from the complete
Riply’s data set riply trn. The data are binary labeled thus the GMM has two Gaus
sians components. The estimated GMM is visualized as: (i) shape of components of the
GMM (see Figure 4.2a) and (ii) contours of the distribution function (see Figure 4.2b)
.
data = load(’riply_trn’); % load labeled (complete) data
model = mlcgmm(data); % ML estimate of GMM
% visualization
figure; hold on;
ppatterns(data); pgauss(model);
figure; hold on;
ppatterns(data);
pgmm(model,struct(’visual’,’contour’));
4.2.2 Incomplete data
The input is a set T
X
= ¦x
1
, . . . , x
l
¦ which contains only vectors of observations
x
i
∈ R
n
without knowledge about the corresponding hidden states y
i
∈ ¸. The
logarithm of probability P(T
X
[θ) is referred to as the loglikelihood L(θ[T
X
) of θ with
respect to T
X
. Thanks to the i.i.d. assumption the loglikelihood can be factorized as
L(θ[T
X
) = log P(T
X
[θ) =
l
i=1
y∈Y
P
XY Θ
(x
i
[y
i
, θ)P
Y Θ
(y
i
[θ) .
45
The problem of MaximumLikelihood (ML) estimation from the incomplete data T
X
is
deﬁned as
θ
∗
= argmax
θ∈Θ
L(θ[T
X
) = argmax
θ∈Θ
l
i=1
y∈Y
P
XY Θ
(x
i
[y
i
, θ)P
Y Θ
(y
i
[θ) . (4.4)
The ML estimate of the parameters θ of the GMM from the incomplete data set
T
XY
does not have an analytical solution. The numerical optimization has to be used
instead.
The ExpectationMaximization (EM) algorithm is an iterative procedure for ML
estimation from the incomplete data. The EM algorithm builds a sequence of parameter
estimates
θ
(0)
, θ
(1)
, . . . , θ
(t)
,
such that the loglikelihood L(θ
(t)
[T
X
) monotonically increases, i.e.,
L(θ
(0)
[T
X
) < L(θ
(1)
[T
X
) < . . . < L(θ
(t)
[T
X
)
until a stationary point L(θ
(t−1)
[T
X
) = L(θ
(t)
[T
X
) is achieved. The EM algorithm is a
local optimization technique thus the estimated parameters depend on the initial guess
θ
(0)
. The random guess or Kmeans algorithm (see Section 4.5) are commonly used to
select the initial θ
(0)
. The EM algorithm for the GMM is implemented in the function
emgmm. An interactive demo on the EM algorithm for the GMM is implemented in
demo emgmm.
References: The EM algorithm (including convergence proofs) is described in book [26].
A nice description can be found in book [3]. The monograph [18] is entirely devoted
to the EM algorithm. The ﬁrst papers describing the EM are [25, 5].
Example: MaximumLikelihood estimate from incomplete data
The estimation of the parameters of the GMM is demonstrated on a synthetic
data. The ground truth model is the GMM with two univariate Gaussian components.
The ground truth model is sampled to obtain a training (incomplete) data. The EM
algorithm is applied to estimate the GMM parameters. The random initialization is
used by default. The ground truth model, sampled data and the estimate model are
visualized in Figure 4.3a. The plot of the loglikelihood function L(θ
(t)
[T
X
) with respect
to the number of iterations t is visualized in Figure 4.3b.
% ground truth Gaussian mixture model
true_model.Mean = [2 2];
true_model.Cov = [1 0.5];
true_model.Prior = [0.4 0.6];
sample = gmmsamp(true_model, 250);
46
% ML estimation by EM
options.ncomp = 2; % number of Gaussian components
options.verb = 1; % display progress info
estimated_model = emgmm(sample.X,options);
% visualization
figure;
ppatterns(sample.X);
h1 = pgmm(true_model,struct(’color’,’r’));
h2 = pgmm(estimated_model,struct(’color’,’b’));
legend([h1(1) h2(1)],’Ground truth’, ’ML estimation’);
figure; hold on;
xlabel(’iterations’); ylabel(’loglikelihood’);
plot( estimated_model.logL );
4.3 Minimax estimation
The input is a set T
X
= ¦x
1
, . . . , x
l
¦ contains vectors x
i
∈ R
n
which are realizations of
random variable distributed according to P
XΘ
(x[θ). The vectors T
X
are assumed to be
a realizations with high value of the distribution function P
XΘ
(x[θ). The P
XΘ
(x[θ)
is known up to the parameter θ ∈ Θ. The minimax estimation is deﬁned as
θ
∗
= argmax
θ∈Θ
min
x∈T
X
log P
XΘ
(x[θ) . (4.5)
If a procedure for the MaximumLikelihood estimate (global solution) of the distri
bution P
XΘ
(x[θ) exists then a numerical iterative algorithm which converges to the
optimal estimate θ
∗
can be constructed. The algorithm builds a series of estimates
θ
(0)
, θ
(1)
, . . . , θ
(t)
which converges to the optimal estimate θ
∗
. The algorithm converges
in a ﬁnite number of iterations to such an estimate θ that the condition
min
x∈T
X
log P
XΘ
(x[θ
∗
) − min
x∈T
X
log P
XΘ
(x[θ) < ε , (4.6)
holds for ε > 0 which deﬁnes closeness to the optimal solution θ
∗
. The inequality (4.6)
can be evaluated eﬃciently thus it is a natural choice for the stopping condition of
the algorithm. The STPRtool contains an implementation of the minimax algorithm
for the Gaussian distribution (4.1). The algorithm is implemented in the function
mmgauss. An interactive demo on the minimax estimation for the Gaussian distribution
is implemented in the function demo mmgauss.
References: The minimax algorithm for probability estimation is described and ana
lyzed in Chapter 8 of the book [26].
47
Example: Minimax estimation of the Gaussian distribution
The bivariate Gaussian distribution is described by three samples which are known
to have high value of the distribution function. The minimax algorithm is applied
to estimate the parameters of the Gaussian. In this particular case, the minimax
problem is equivalent to searching for the minimal enclosing ellipsoid. The found
solution is visualized as the isoline of the distribution function at the point P
XΘ
(x[θ)
(see Figure 4.4).
X = [[0;0] [1;0] [0;1]]; % points with high p.d.f.
model = mmgauss(X); % run minimax algorithm
% visualization
figure; ppatterns(X,’xr’,13);
pgauss(model, struct(’p’,exp(model.lower_bound)));
4.4 Probabilistic output of classiﬁer
In general, the discriminant function of an arbitrary classiﬁer does not have meaning of
probability (e.g., SVM, Perceptron). However, the probabilistic output of the classiﬁer
can help in postprocessing, for instance, in combining more classiﬁers together. Fitting
a sigmoid function to the classiﬁer output is a way how to solve this problem.
Let T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ is the training set composed of the vectors x
i
∈
A ⊆ R
n
and the corresponding binary hidden states y
i
∈ ¸ = ¦1, 2¦. The training
set T
XY
is assumed to be identically and independently distributed from underlying
distribution. Let f: A ⊆ R
n
→R be a discriminant function trained from the data T
XY
by an arbitrary learning method. The aim is to estimate parameters of a posteriori
distribution P
Y FΘ
(y[f(x), θ) of the hidden state y given the value of the discriminant
function f(x). The distribution is modeled by a sigmoid function
P
Y FΘ
(1[f(x), θ) =
1
1 + exp(a
1
f(x) + a
2
)
,
which is determined by parameters θ = [a
1
, a
2
]
T
. The loglikelihood function is deﬁned
as
L(θ[T
XY
) =
l
i=1
log P
Y FΘ
(y
i
[f(x
i
), θ) .
The parameters θ = [a
1
, a
2
]
T
can be estimated by the maximumlikelihood method
θ = argmax
θ
L(θ
[T
XY
) = argmax
θ
l
i=1
log P
Y FΘ
(y
i
[f(x
i
), θ
) . (4.7)
48
The STPRtool provides an implementation mlsigmoid which uses the Matlab Opti
mization toolbox function fminunc to solve the task (4.7). The function produces the
parameters θ = [a
1
, a
2
]
T
which are stored in the datatype describing the sigmoid (see
Table 4.6). The evaluation of the sigmoid is implemented in function sigmoid.
Table 4.6: Datatype used to describe sigmoid function.
Sigmoid function (structure array):
.A [2 1] Parameters θ = [a
1
, a
2
]
T
of the sigmoid function.
.fun = ’sigmoid’ Identiﬁes function associated with this data type.
References: The method of ﬁtting the sigmoid function is described in [22].
Example: Probabilistic output for SVM
This example is implemented in the script demo svmpout. The example shows how
to train the probabilistic output for the SVM classiﬁer. The Riply’s data set riply trn
is used for training the SVM and the ML estimation of the sigmoid. The ML estimated
of Gaussian mixture model (GMM) of the SVM output is used for comparison. The
GMM allows to compute the a posteriori probability for comparison to the sigmoid
estimate. The example is implemented in the script demo svmpout. Figure 4.5a shows
found probabilistic models of a posteriori probability of the SVM output. Figure 4.5b
shows the decision boundary of SVM classiﬁer for illustration.
% load training data
data = load(’riply_trn’);
% train SVM
options = struct(’ker’,’rbf’,’arg’,1,’C’,10);
svm_model = smo(data,options);
% compute SVM output
[dummy,svm_output.X] = svmclass(data.X,svm_model);
svm_output.y = data.y;
% fit sigmoid function to svm output
sigmoid_model = mlsigmoid(svm_output);
% ML estimation of GMM model of SVM output
gmm_model = mlcgmm(svm_output);
% visulization
figure; hold on;
49
xlabel(’svm output f(x)’); ylabel(’p(y=1f(x))’);
% sigmoid model
fx = linspace(min(svm_output.X), max(svm_output.X), 200);
sigmoid_apost = sigmoid(fx,sigmoid_model);
hsigmoid = plot(fx,sigmoid_apost,’k’);
ppatterns(svm_output);
% GMM model
pcond = pdfgauss( fx, gmm_model);
gmm_apost = (pcond(1,:)*gmm_model.Prior(1))./...
(pcond(1,:)*gmm_model.Prior(1)+(pcond(2,:)*gmm_model.Prior(2)));
hgmm = plot(fx,gmm_apost,’g’);
hcomp = pgauss(gmm_model);
legend([hsigmoid,hgmm,hcomp],’P(y=1f(x)) MLSigmoid’,...
’P(y=1f(x)) MLGMM’,’P(f(x)y=1) MLGMM’,’P(f(x)y=2) MLGMM’);
% SVM decision boundary
figure; ppatterns(data); psvm(svm_model);
4.5 Kmeans clustering
The input is a set T
X
= ¦x
1
, . . . , x
l
¦ which contains vectors x
i
∈ R
n
. Let θ =
¦µ
1
, . . . , µ
c
¦, µ
i
∈ R
n
be a set of unknown cluster centers. Let the objective function
F(θ) be deﬁned as
F(θ) =
1
l
l
i=1
min
y=1,...,c
µ
y
−x
i

2
.
The small value of F(θ) indicates that the centers θ well describe the input set T
X
.
The task is to ﬁnd such centers θ which describe the set T
X
best, i.e., the task is to
solve
θ
∗
= argmin
θ
F(θ) = argmin
θ
1
l
l
i=1
min
y=1,...,c
µ
y
−x
i

2
.
The Kmeans
1
is an iterative procedure which iteratively builds a series θ
(0)
, θ
(1)
, . . . , θ
(t)
.
The Kmeans algorithm converges in a ﬁnite number of steps to the local optimum of
the objective function F(θ). The found solution depends on the initial guess θ
(0)
which
1
The K in the name Kmeans stands for the number of centers which is here denoted by c.
50
is usually selected randomly. The Kmeans algorithm is implemented in the function
kmeans.
References: The Kmeans algorithm is described in the books [26, 6, 3].
Example: Kmeans clustering
The Kmeans algorithm is used to ﬁnd 4 clusters in the Riply’s training set. The
solution is visualized as the cluster centers. The vectors classiﬁed to the corresponding
centers are distinguished by color (see Figure 4.6a). The plot of the objective function
F(θ
(t)
) with respect to the number of iterations is shown in Figure 4.6b.
data = load(’riply_trn’); % load data
[model,data.y] = kmeans(data.X, 4 ); % run kmeans
% visualization
figure; ppatterns(data);
ppatterns(model.X,’sk’,14);
pboundary( model );
figure; hold on;
xlabel(’t’); ylabel(’F’);
plot(model.MsErr);
51
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
gauss 1
0.981
gauss 2
1.45
(a)
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
(b)
Figure 4.2: Example of the MaximumLikelihood of the Gaussian Mixture Model from
the incomplete data. Figure (a) shows components of the GMM and Figure (b) shows
contour plot of the distribution function.
52
−5 −4 −3 −2 −1 0 1 2 3 4 5
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Ground truth
ML estimation
(a)
0 5 10 15 20 25
−460
−458
−456
−454
−452
−450
−448
−446
−444
−442
iterations
l
o
g
−
l
i
k
e
l
i
h
o
o
d
(b)
Figure 4.3: Example of using the EM algorithm for ML estimation of the Gaussian
mixture model. Figure (a) shows ground truth and the estimated GMM and Figure(b)
shows the loglikelihood with respect to the number of iterations of the EM.
53
−0.4 −0.2 0 0.2 0.4 0.6 0.8 1
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
gauss 1
0.304
Figure 4.4: Example of the Minimax estimation of the Gaussian distribution.
54
−4 −3 −2 −1 0 1 2 3 4 5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
svm output f(x)
p
(
y
=
1

f
(
x
)
)
P(y=1f(x)) ML−Sigmoid
P(y=1f(x)) ML−GMM
P(f(x)y=1) ML−GMM
P(f(x)y=2) ML−GMM
(a)
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
(b)
Figure 4.5: Example of ﬁtting a posteriori probability to the SVM output. Figure (a)
shows the sigmoid model and GMM model both ﬁtted by ML estimated. Figure (b)
shows the corresponding SVM classiﬁer.
55
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
(a)
0 2 4 6 8 10 12
0.04
0.06
0.08
0.1
0.12
0.14
0.16
t
F
(b)
Figure 4.6: Example of using the Kmeans algorithm to cluster the Riply’s data set.
Figure (a) shows found clusters and Figure (b) contains the plot of the objective func
tion F(θ
(t)
) with respect to the number of iterations.
56
Chapter 5
Support vector and other kernel
machines
5.1 Support vector classiﬁers
The binary support vector classiﬁer uses the discriminant function f: A ⊆ R
n
→R of
the following form
f(x) = ¸α k
S
(x)) + b . (5.1)
The k
S
(x) = [k(x, s
1
), . . . , k(x, s
d
)]
T
is the vector of evaluations of kernel functions
centered at the support vectors o = ¦s
1
, . . . , s
d
¦, s
i
∈ R
n
which are usually subset of
the training data. The α ∈ R
l
is a weight vector and b ∈ R is a bias. The binary
classiﬁcation rule q: A → ¸ = ¦1, 2¦ is deﬁned as
q(x) =
_
1 for f(x) ≥ 0 ,
2 for f(x) < 0 .
(5.2)
The datatype used by the STPRtool to represent the binary SVM classiﬁer is described
in Table 5.2.
The multiclass generalization involves a set of discriminant functions f
y
: A ⊆ R
n
→
R, y ∈ ¸ = ¦1, 2, . . . , c¦ deﬁned as
f
y
(x) = ¸α
y
k
S
(x)) + b
y
, y ∈ ¸ .
Let the matrix A = [α
1
, . . . , α
c
] be composed of all weight vectors and b = [b
1
, . . . , b
c
]
T
be a vector of all biases. The multiclass classiﬁcation rule q: A → ¸ = ¦1, 2, . . . , c¦ is
deﬁned as
q(x) = argmax
y∈Y
f
y
(x) . (5.3)
The datatype used by the STPRtool to represent the multiclass SVM classiﬁer is
described in Table 5.3. Both the binary and the multiclass SVM classiﬁers are imple
mented in the function svmclass.
57
The majority voting strategy is other commonly used method to implement the
multiclass SVM classiﬁer. Let q
j
: A ⊆ R
n
→ ¦y
1
j
, y
2
j
¦, j = 1, . . . , g be a set of g binary
SVM rules (5.2). The jth rule q
j
(x) classiﬁes the inputs x into class y
1
j
∈ ¸ or y
2
j
∈ ¸.
Let v(x) be a vector [c 1] of votes for classes ¸ when the input x is to be classiﬁed.
The vector v(x) = [v
1
(x), . . . , v
c
(x)]
T
is computed as:
Set v
y
= 0, ∀y ∈ ¸.
For j = 1, . . . , g do
v
y
(x) = v
y
(x) + 1 where y = q
j
(x).
end.
The majority votes based multiclass classiﬁer assigns the input x into such class y ∈ ¸
having the majority of votes
y = argmax
y
=1,...,g
v
y
(x) . (5.4)
The datatype used to represent the majority voting strategy for the multiclass SVM
classiﬁer is described in Table 5.4. The strategy (5.4) is implemented in the function
mvsvmclass. The implemented methods to deal with the SVM and related kernel
classiﬁers are enlisted in Table 5.1
References: Books describing Support Vector Machines are for example [31, 4, 30]
5.2 Binary Support Vector Machines
The input is a set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of training vectors x
i
∈ R
n
and cor
responding hidden states y
i
∈ ¸ = ¦1, 2¦. The linear SVM aims to train the linear
discriminant function f(x) = ¸w x) + b of the binary classiﬁer
q(x) =
_
1 for f(x) ≥ 0 ,
2 for f(x) < 0 .
The training of the optimal parameters (w
∗
, b
∗
) is transformed to the following quadratic
programming task
(w
∗
, b
∗
) = argmin
w,b
1
2
w
2
+ C
l
i=1
ξ
p
i
, (5.5)
¸w x
i
) + b ≥ +1 −ξ
i
, i ∈ 1
1
,
¸w x
i
) + b ≤ −1 + ξ
i
, i ∈ 1
2
,
ξ
i
≥ 0 , i ∈ 1
1
∪ 1
2
.
58
Table 5.1: Implemented methods: Support vector and other kernel machines.
svmclass Support Vector Machines Classiﬁer.
mvsvmclass Majority voting multiclass SVM classiﬁer.
evalsvm Trains and evaluates Support Vector Machines classi
ﬁer.
bsvm2 Multiclass BSVM with L2soft margin.
oaasvm Multiclass SVM using OneAgainstAll decomposi
tion.
oaosvm Multiclass SVM using OneAgainstOne decomposi
tion.
smo Sequential Minimal Optimization for binary SVM
with L1soft margin.
svmlight Interface to SV M
light
software.
svmquadprog SVM trained by Matlab Optimization Toolbox.
diagker Returns diagonal of kernel matrix of given data.
kdist Computes distance between vectors in kernel space.
kernel Evaluates kernel function.
kfd Kernel Fisher Discriminant.
kperceptr Kernel Perceptron.
minball Minimal enclosing ball in kernel feature space.
rsrbf Reduced Set Method for RBF kernel expansion.
rbfpreimg RBF preimage by Schoelkopf’s ﬁxedpoint algorithm.
rbfpreimg2 RBF preimage problem by Gradient optimization.
rbfpreimg3 RBF preimage problem by KwokTsang’s algorithm.
demo svm Demo on Support Vector Machines.
The 1
1
= ¦i: y
i
= 1¦ and 1
2
= ¦i: y
i
= 2¦ are sets of indices. The C > 0 stands for the
regularization constant. The ξ
i
≥ 0, i ∈ 1
1
∪ 1
2
are slack variables used to relax the
inequalities for the case of nonseparable data. The linear p = 1 and quadratic p = 2
term for the slack variables ξ
i
is used. In the case of the parameter p = 1, the L1soft
margin SVM is trained and for p = 2 it is denoted as the L2soft margin SVM.
The training of the nonlinear (kernel) SVM with L1soft margin corresponds to
solving the following QP task
β
∗
= argmax
β
¸β 1) −
1
2
¸β Hβ) , (5.6)
subject to
¸β γ) = 1 ,
β ≥ 0 ,
β ≤ 1C .
59
Table 5.2: Datatype used to describe binary SVM classiﬁer.
Binary SVM classiﬁer (structure array):
.Alpha [d 1] Weight vector α.
.b [1 1] Bias b.
.sv.X [n d] Support vectors o = ¦s
1
, . . . , s
d
¦.
.options.ker [string] Kernel identiﬁer.
.options.arg [1 p] Kernel argument(s).
.fun = ’svmclass’ Identiﬁes function associated with this data type.
Table 5.3: Datatype used to describe multiclass SVM classiﬁer.
Multiclass SVM classiﬁer (structure array):
.Alpha [d c] Weight vectors A = [α
1
, . . . , α
c
].
.b [c 1] Vector of biased b = [b
1
, . . . , b
c
]
T
.
.sv.X [n d] Support vectors o = ¦s
1
, . . . , s
d
¦.
.options.ker [string] Kernel identiﬁer.
.options.arg [1 p] Kernel argument(s).
.fun = ’svmclass’ Identiﬁes function associated with this data type.
The 0 [l 1] is vector of zeros and the 1 [l 1] is vector of ones. The vector γ [l 1]
and the matrix H [l l] are constructed as follows
γ
i
=
_
1 if y
i
= 1 ,
−1 if y
i
= 2 ,
H
i,j
=
_
k(x
i
, x
j
) if y
i
= y
j
,
−k(x
i
, x
j
) if y
i
,= y
j
,
where k: A A → R is selected kernel function. The discriminant function (5.1) is
determined by the weight vector α, bias b and the set of support vectors o. The
set of support vectors is denoted as o = ¦x
i
: i ∈ 1
SV
¦, where the set of indices is
1
SV
= ¦i: β
i
> 0¦. The weight vectors are compute as
α
i
=
_
β
i
if y
i
= 1 ,
−β
i
if y
i
= 2 ,
for i ∈ 1
SV
. (5.7)
The bias b can be computed from the KarushKuhnTucker (KKT) conditions as the
following constrains
f(x
i
) = ¸α k
S
(x
i
)) + b = +1 for i ∈ ¦j: y
j
= 1, 0 < β
j
< C¦ ,
f(x
i
) = ¸α k
S
(x
i
)) + b = −1 for i ∈ ¦j: y
j
= 2, 0 < β
j
< C¦ ,
must hold for the optimal solution. The bias b is computed as an average over all the
constrains such that
b =
1
[1
b
[
i∈I
b
γ
i
−¸α k
S
(x
i
)) .
60
Table 5.4: Datatype used to describe the majority voting strategy for multiclass SVM
classiﬁer.
Majority voting SVM classiﬁer (structure array):
.Alpha [d g] Weight vectors A = [α
1
, . . . , α
g
].
.b [g 1] Vector of biased b = [b
1
, . . . , b
g
]
T
.
.bin y [2 g] Targets for the binary rules. The vector bin y(1,:)
contains [y
1
1
, y
1
2
, . . . , y
1
g
] and the bin y(2,:) contains
[y
2
1
, y
2
2
, . . . , y
2
g
].
.sv.X [n d] Support vectors o = ¦s
1
, . . . , s
d
¦.
.options.ker [string] Kernel identiﬁer.
.options.arg [1 p] Kernel argument(s).
.fun = ’mvsvmclass’ Identiﬁes function associated with this data type.
where 1
b
= ¦i: 0 < β
i
< C¦ are the indices of the vectors on the boundary.
The training of the nonlinear (kernel) SVM with L2soft margin corresponds to
solving the following QP task
β
∗
= argmax
β
¸β 1) −
1
2
¸β (H+
1
2C
E)β) , (5.8)
subject to
¸β γ) = 1 ,
β ≥ 0 ,
where E is the identity matrix. The set of support vectors is set to o = ¦x
i
: i ∈ 1
SV
¦
where the set of indices is 1
SV
= ¦i: β
i
> 0¦. The weight vector α is given by (5.7).
The bias b can be computed from the KKT conditions as the following constrains
f(x
i
) = ¸α k
S
(x
i
)) + b = +1 −
α
i
2C
for i ∈ ¦i: y
j
= 1, β
j
> 0¦ ,
f(x
i
) = ¸α k
S
(x
i
)) + b = −1 +
α
i
2C
for i ∈ ¦i: y
j
= 2, β
j
> 0¦ ,
must hold at the optimal solution. The bias b is computed as an average over all the
constrains thus
b =
1
[1
SV
[
i∈I
SV
γ
i
(1 −
α
i
2C
) −¸α k
S
(x
i
)) .
The binary SVM solvers implemented in the STPRtool are enlisted in Table 5.5.
An interactive demo on the binary SVM is implemented in demo svm.
References: The Sequential Minimal Optimizer (SMO) is described in the books [23,
27, 30]. The SV M
light
is described for instance in the book [27].
Example: Training binary SVM.
61
Table 5.5: Implemented binary SVM solvers.
Function description
smo Implementation of the Sequential Minimal Optimizer (SMO)
used to train the binary SVM with L1soft margin. It can
handle moderate size problems.
svmlight Matlab interface to SV M
light
software (Version: 5.00) which
trains the binary SVM with L1soft margin. The executable
ﬁle svm learn for the Linux OS must be installed. It can be
downloaded from http://svmlight.joachims.org/. The
SV M
light
is very powerful optimizer which can handle large
problems.
svmquadprog SVM solver based on the QP solver quadprog of the Mat
lab optimization toolbox. It trains both L1soft and L2soft
margin binary SVM. The function quadprog is a part of the
Matlab Optimization toolbox. It useful for small problems
only.
The binary SVM is trained on the Riply’s training set riply trn. The SMO solver
is used in the example but other solvers (svmlight, svmquadprog) can by used as well.
The trained SVM classiﬁer is evaluated on the testing data riply tst. The decision
boundary is visualized in Figure 5.1.
trn = load(’riply_trn’); % load training data
options.ker = ’rbf’; % use RBF kernel
options.arg = 1; % kernel argument
options.C = 10; % regularization constant
% train SVM classifier
model = smo(trn,options);
% model = svmlight(trn,options);
% model = svmquadprog(trn,options);
% visualization
figure;
ppatterns(trn); psvm(model);
tst = load(’riply_tst’); % load testing data
ypred = svmclass(tst.X,model); % classify data
cerror(ypred,tst.y) % compute error
62
ans =
0.0920
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.1: Example showing the binary SVM classiﬁer trained on the Riply’s data.
5.3 OneAgainstAll decomposition for SVM
The input is a set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of training vectors x
i
∈ A ⊆ R
n
and corresponding hidden states y
i
∈ ¸ = ¦1, 2, . . . , c¦. The goal is to train the
multiclass rule q: A ⊆ R
n
→ ¸ deﬁned by (5.3). The problem can be solved by
the OneAgainstAll (OAA) decomposition. The OAA decomposition transforms the
multiclass problem into a series of c binary subtasks that can be trained by the binary
SVM. Let the training set T
y
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ contain the modiﬁed hidden
states deﬁned as
y
i
=
_
1 for y = y
i
,
2 for y ,= y
i
.
The discriminant functions
f
y
(x) = ¸α
y
k
S
(x)) + b
y
, y ∈ ¸ ,
are trained by the binary SVM solver from the set T
y
XY
, y ∈ ¸. The OAA decomposition
is implemented in the function oaasvm. The function allows to specify any binary SVM
63
solver (see Section 5.2) used to solve the binary problems. The result is the multiclass
SVM classiﬁer (5.3) represented by the datatype described in Table 5.3.
References: The OAA decomposition to train the multiclass SVM and comparison
to other methods is described for instance in the paper [12].
Example: Multiclass SVM using the OAA decomposition
The example shows the training of the multiclass SVM using the OAA decomposi
tion. The SMO binary solver is used to train the binary SVM subtasks. The training
data and the decision boundary is visualized in Figure 5.2.
data = load(’pentagon’); % load data
options.solver = ’smo’; % use SMO solver
options.ker = ’rbf’; % use RBF kernel
options.arg = 1; % kernel argument
options.C = 10; % regularization constant
model = oaasvm(data,options ); % training
% visualization
figure;
ppatterns(data);
ppatterns(model.sv.X,’ko’,12);
pboundary(model);
5.4 OneAgainstOne decomposition for SVM
The input is a set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of training vectors x
i
∈ A ⊆ R
n
and
corresponding hidden states y
i
∈ ¸ = ¦1, 2, . . . , c¦. The goal is to train the multiclass
rule q: A ⊆ R
n
→ ¸ based on the majority voting strategy deﬁned by (5.4). The
oneagainstone (OAO) decomposition strategy can be used train such a classiﬁer. The
OAO decomposition transforms the multiclass problem into a series of g = c(c −1)/2
binary subtasks that can be trained by the binary SVM. Let the training set T
j
XY
=
¦(x
1
, y
1
), . . . , (x
l
j
, y
l
j
)¦ contain the training vectors x
i
∈ 1
j
= ¦i: y
i
= y
1
_
y
i
= y
2
¦
and the modiﬁed the hidden states deﬁned as
y
i
=
_
1 for y
1
j
= y
i
,
2 for y
2
j
= y
i
,
i ∈ 1
j
.
The training set T
j
XY
, j = 1, . . . , g is constructed for all g = c(c −1)/2 combinations of
classes y
1
j
∈ ¸ and y
2
j
∈ ¸ ¸¦y
1
j
¦. The binary SVM rules q
j
, j = 1, . . . , g are trained on
the data T
j
XY
. The OAO decomposition for multiclass SVM is implemented in function
64
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Figure 5.2: Example showing the multiclass SVM classiﬁer build by oneagainstall
decomposition.
oaosvm. The function oaosvm allows to specify an arbitrary binary SVM solver to train
the subtasks.
References: The OAO decomposition to train the multiclass SVM and comparison
to other methods is described for instance in the paper [12].
Example: Multiclass SVM using the OAO decomposition
The example shows the training of the multiclass SVM using the OAO decomposi
tion. The SMO binary solver is used to train the binary SVM subtasks. The training
data and the decision boundary is visualized in Figure 5.3.
data = load(’pentagon’); % load data
options.solver = ’smo’; % use SMO solver
options.ker = ’rbf’; % use RBF kernel
options.arg = 1; % kernel argument
options.C = 10; % regularization constant
model = oaosvm(data,options); % training
% visualization
figure;
ppatterns(data);
ppatterns(model.sv.X,’ko’,12);
pboundary(model);
65
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Figure 5.3: Example showing the multiclass SVM classiﬁer build by oneagainstone
decomposition.
5.5 Multiclass BSVM formulation
The input is a set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of training vectors x
i
∈ A ⊆ R
n
and
corresponding hidden states y
i
∈ ¸ = ¦1, 2, . . . , c¦. The goal is to train the multiclass
rule q: A ⊆ R
n
→ ¸ deﬁned by (5.3). In the linear case, the parameters of the multi
class rule can be found by solving the multiclass BSVM formulation which is deﬁned
as
(W
∗
, b
∗
, ξ
∗
) = argmin
W,b
1
2
y∈Y
(w
y

2
+ b
2
y
) + C
i∈I
y∈Y\{y}
(ξ
i
)
p
. ¸¸ .
F(W,b,ξ)
, (5.9)
subject to
¸w
y
i
x
i
) + b
y
i
−(¸w
y
x
i
) + b
y
) ≥ 1 −ξ
y
i
, i ∈ 1 , y ∈ ¸ ¸ ¦y
i
¦ ,
ξ
y
i
≥ 0 , i ∈ 1 , y ∈ ¸ ¸ ¦y
i
¦ ,
where W = [w
1
, . . . , w
c
] is a matrix of normal vectors, b = [b
1
, . . . , b
c
]
T
is vector of
biases, ξ is a vector of slack variables and 1 = [1, . . . , l] is set of indices. The L1soft
margin is used for p = 1 and L2soft margin for p = 2. The letter B in BSVM stands
for the bias term added to the objective (5.9).
In the case of the L2soft margin p = 2, the optimization problem (5.9) can be
66
expressed in its dual form
A
∗
= argmax
A
i∈I
y∈Y\{y
i
}
α
y
i
−
1
2
i∈I
y∈Y\{y
i
}
j∈I
u∈Y\{y
j
}
α
y
i
α
u
j
h(y, u, i, j) , (5.10)
subject to
α
y
i
≥ 0 , i ∈ 1 , y ∈ ¸ ¸ ¦y
i
¦ ,
where A = [α
1
, . . . , α
c
] are optimized weight vectors. The function h: ¸¸11 →R
is deﬁned as
h(y, u, i, j) = (¸x
i
x
j
)+1)(δ(y
i
, y
j
)+δ(y, u)−δ(y
i
, u)−δ(y
j
, y))+
δ(y, u)δ(i, j)
2C
. (5.11)
The criterion (5.10) corresponds to the singleclass SVM criterion which is simple to
optimize. The multiclass rule (5.3) is composed of the set of discriminant functions
f
y
: A →R which are computed as
f
y
(x) =
i∈I
¸x
i
x)
u∈Y\{y
i
}
α
u
i
(δ(u, y
i
) −δ(u, y)) + b
y
, y ∈ ¸ , (5.12)
where the bias b
y
, y ∈ ¸ is given by
b
y
=
i∈I
y∈Y\{y
i
}
α
u
i
(δ(y, y
i
) −δ(y, u)) , y ∈ ¸ .
The nonlinear (kernel) classiﬁer is obtained by substituting the selected kernel k: A
A →R for the dot products ¸xx
) to (5.11) and (5.12). The training of the multiclass
BSVM classiﬁer with L2soft margin is implemented in the function bsvm2. The opti
mization task (5.10) can be solved by most the optimizers. The optimizers implemented
in the function bsvm2 are enlisted bellow:
MitchellDemyanovMalozemov (solver = ’mdm’).
Kozinec’s algorithm (solver = ’kozinec’).
Nearest Point Algorithm (solver = ’npa’).
The above mentioned algorithms are of iterative nature and converge to the optimal
solution of the criterion (5.9). The upper bound F
UB
and the lower bound F
LB
on the
optimal value F(W
∗
, b
∗
, ξ
∗
) of the criterion (5.9) can be computed. The bounds are
used in the algorithms to deﬁne the following two stopping conditions:
Relative tolerance:
F
UB
−F
LB
F
LB
+ 1
≤ ε
rel
.
67
Absolute tolerance: F
UB
≤ ε
abs
.
References: The multiclass BSVM formulation is described in paper [12]. The trans
formation of the multiclass BSVM criterion to the singleclass criterion (5.10) imple
mented in the STPRtool is published in paper [9]. The MitchellDemyanovMalozemov
and Nearest Point Algorithm are described in paper [14]. The Kozinec’s algorithm
based SVM solver is described in paper [10].
Example: Multiclass BSVM with L2soft margin.
The example shows the training of the multiclass BSVM using the NPA solver. The
other solvers can be used instead by setting the variable options.solver (other options
are ’kozinec’, ’mdm’). The training data and the decision boundary is visualized in
Figure 5.4.
data = load(’pentagon’); % load data
options.solver = ’npa’; % use NPA solver
options.ker = ’rbf’; % use RBF kernel
options.arg = 1; % kernel argument
options.C = 10; % regularization constant
model = bsvm2(data,options); % training
% visualization
figure;
ppatterns(data);
ppatterns(model.sv.X,’ko’,12);
pboundary(model);
5.6 Kernel Fisher Discriminant
The input is a set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ of training vectors x
i
∈ A ⊆ R
n
and
corresponding values of binary state y
i
∈ ¸ = ¦1, 2¦. The Kernel Fisher Discriminant
is the nonlinear extension of the linear FLD (see Section 2.3). The input training
vectors are assumed to be mapped into a new feature space T by a mapping function
φ: A → T. The ordinary FLD is applied on the mapped data training data T
ΦY
=
¦(φ(x
1
), y
1
), . . . , (φ(x
l
), y
l
)¦. The mapping is performed eﬃciently by means of kernel
functions k: A A → R being the dot products of the mapped vectors k(x, x
) =
¸φ(x) φ(x
)). The aim is to ﬁnd a direction ψ =
l
i=1
α
i
φ(x
i
) in the feature space
T given by weights α = [α
1
, . . . , α
l
]
T
such that the criterion
F(α) =
¸α Mα)
¸α (N+ µE)α)
=
¸α m)
2
¸α (N+ µE)α)
, (5.13)
68
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Figure 5.4: Example showing the multiclass BSVM classiﬁer with L2soft margin.
is maximized. The criterion (5.13) is the feature space counterpart of the crite
rion (2.13) deﬁned in the input space A. The vector m [l 1], matrices N [l l]
M [l l] are constructed as
m
y
=
1
[1
y
[
K1
y
, y ∈ ¸ , m = m
1
−m
2
,
N = KK
T
−
y∈Y
[1
y
[m
y
m
T
y
, M = (m
1
−m
2
)(m
1
−m
2
)
T
,
where the vector 1
y
has entries i ∈ 1
y
equal to and remaining entries zeros. The kernel
matrix K [l l] contains kernel evaluations K
i,j
= k(x
i
, x
j
). The diagonal matrix
µE in the denominator of the criterion (5.13) serves as the regularization term. The
discriminant function f(x) of the binary classiﬁer (5.2) is given by the found vector α
as
f(x) = ¸ψ φ(x)) + b = ¸α k(x)) + b , (5.14)
where the k(x) = [k(x
1
, x), . . . , k(x
l
, x)]
T
is vector of evaluations of the kernel. The
discriminant function (5.14) is known up to the bias b which is determined by the linear
SVM with L1soft margin is applied on the projected data T
ZY
= ¦(z
1
, y
1
), . . . , (z
l
, y
l
)¦.
The projections ¦z
1
, . . . , z
l
¦ are computed as
z
i
= ¸α k(x
i
)) .
The KFD training algorithm is implemented in the function kfd.
69
References: The KFD formulation is described in paper [19]. The detailed analysis
can be found in book [30].
Example: Kernel Fisher Discriminant
The binary classiﬁer is trained by the KFD on the Riply’s training set riply trn.mat.
The found classiﬁer is evaluated on the testing data riply tst.mat. The training data
and the found kernel classiﬁer is visualized in Figure 5.5. The classiﬁer is visualized as
the SVM classiﬁer, i.e., the isolines with f(x) = +1 and f(x) = −1 are visualized and
the support vectors (whole training set) are marked as well.
trn = load(’riply_trn’); % load training data
options.ker = ’rbf’; % use RBF kernel
options.arg = 1; % kernel argument
options.C = 10; % regularization of linear SVM
options.mu = 0.001; % regularization of KFD criterion
model = kfd(trn, options) % KFD training
% visualization
figure;
ppatterns(trn);
psvm(model);
% evaluation on testing data
tst = load(’riply_tst’);
ypred = svmclass( tst.X, model );
cerror( ypred, tst.y )
ans =
0.0960
5.7 Preimage problem
Let T
X
= ¦x
1
, . . . , x
l
¦ be a set of vector from the input space A ⊆ R
n
. The kernel
methods works with the data nonlinearly mapped φ: A → T into a new feature space
T. The mapping is performed implicitly by using the kernel functions k: A A → R
which are dot products of the input data mapped into the feature space T such that
k(x
a
, x
b
) = ¸φ(x
a
) φ(x
b
)). Let ψ ∈ T be given by the following kernel expansion
ψ =
l
i=1
α
i
φ(x
i
) , (5.15)
70
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.5: Example shows the binary classiﬁer trained based on the Kernel Fisher
Discriminant.
where α = [α
1
, . . . , α
l
]
T
is a weight vector of real coeﬃcients. The preimage problem
aims to ﬁnd an input vector x ∈ A, its feature space image φ(x) ∈ T would well
approximate the expansion point Ψ ∈ T given by (5.15). In other words, the pre
image problem solves mapping from the feature space T back to the input space A.
The preimage problem can be stated as the following optimization task
x = argmin
x
φ(x) −ψ
2
= argmin
x
φ(x
) −
l
i=1
α
i
φ(x
i
)
2
= argmin
x
k(x
, x
) −2
l
i=1
α
i
k(x
, x
i
) +
l
i=1
l
j=1
α
i
α
j
k(x
i
, x
j
) .
(5.16)
Let the Radial Basis Function (RBF) k(x
a
, x
b
) = exp(−0.5x
a
−x
b

2
/σ
2
) be assumed.
The preimage problem (5.16) for the particular case of the RBF kernel leads to the
following task
x = argmax
x
l
i=1
α
i
exp(−0.5x
−x
i

2
/σ
2
) . (5.17)
The STPRtool provides three optimization algorithms to solve the RBF preimage
problem (5.17) which are enlisted in Table 5.6. However, all the methods are guaranteed
to ﬁnd only a local optimum of the task (5.17).
71
Table 5.6: Methods to solve the RBF preimage problem implemented in the STPRtool.
rbfpreimg An iterative ﬁxed point algorithm proposed in [28].
rbfpreimg2 A standard gradient optimization method using the
Matlab optimization toolbox for 1D search along the
gradient.
rbfpreimg3 A method exploiting the correspondence between dis
tances in the input space and the feature space pro
posed in [16].
References: The preimage problem and the implemented algorithms for its solution
are described in [28, 16].
Example: Preimage problem
The example shows how to visualize a preimage of the data mean in the feature
induced by the RBF kernel. The result is visualized in Figure 5.6.
% load input data
data = load(’riply_trn’);
% define the mean in the feature space
num_data = size(data,2);
expansion.Alpha = ones(num_data,1)/num_data;
expansion.sv.X = data.X;
expansion.options.ker = ’rbf’;
expansion.options.arg = 1;
% compute preimage problem
x = rbfpreimg(expansion);
% visualization
figure;
ppatterns(data.X);
ppatterns(x,’or’,13);
72
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.6: Example shows training data and the preimage of their mean in the feature
space induced by the RBF kernel.
5.8 Reduced set method
The reduced set method is useful to simplify the kernel classiﬁer trained by the SVM
or other kernel machines. Let the function f: A ⊆ R
n
→R be given as
f(x) =
l
i=1
α
i
k(x
i
, x) + b = ¸ψ φ(x)) + b , (5.18)
where T
X
= ¦x
1
, . . . , x
l
¦ are vectors from R
n
, k: A A → R is a kernel function,
α = [α
1
, . . . , α
l
]
T
is a weight vector of real coeﬃcients and b ∈ R is a scalar. Let
φ: A → T be a mapping from the input space A to the feature space T associated
with kernel function such that k(x
a
, x
b
) = ¸φ(x
a
) φ(x
b
)). The function f(x) can be
expressed in the feature space T as the linear function f(x) = ¸ψ φ(x)) + b. The
reduced set method aims to ﬁnd a new function
˜
f(x) =
m
i=1
β
i
k(s
i
, x) + b = ¸φ(x)
m
i=1
β
i
φ(s
i
)
. ¸¸ .
˜
ψ
) + b , (5.19)
such that the expansion (5.19) is shorter, i.e., m < l, and well approximates the original
one (5.18). The weight vector β = [β
1
, . . . , β
m
]
T
and the vectors T
S
= ¦s
1
, . . . , s
m
¦,
s
i
∈ R
n
determine the reduced kernel expansion (5.19). The problem of ﬁnding the
73
reduced kernel expansion (5.19) can be stated as the optimization task
(β, T
S
) = argmin
β
,T
S

˜
ψ −ψ
2
= argmin
β
,T
S
_
_
_
_
_
m
i=1
β
i
φ(s
i
) −
l
i=1
α
i
φ(x
i
)
_
_
_
_
_
2
. (5.20)
Let the Radial Basis Function (RBF) k(x
a
, x
b
) = exp(−0.5x
a
−x
b

2
/σ
2
) be assumed.
In this case, the optimization task (5.20) can be solved by an iterative greedy algo
rithm. The algorithm converts the task (5.20) into a series of m preimage tasks (see
Section 5.7). The reduced set method for the RBF kernel is implemented in function
rsrbf.
References: The implemented reduced set method is described in [28].
Example: Reduced set method for SVM classiﬁer
The example shows how to use the reduced set method to simplify the SVM clas
siﬁer. The SVM classiﬁer is trained from the Riply’s training set riply trn/mat. The
trained kernel expansion (discriminant function) is composed of 94 support vectors.
The reduced expansion with 10 support vectors is found. Decision boundaries of the
original and the reduced SVM are visualized in Figure 5.7. The original and reduced
SVM are evaluated on the testing data riply tst.mat.
% load training data
trn = load(’riply_trn’);
% train SVM classifier
model = smo(trn,struct(’ker’,’rbf’,’arg’,1,’C’,10));
% compute the reduced rule with 10 support vectors
red_model = rsrbf(model,struct(’nsv’,10));
% visualize decision boundaries
figure; ppatterns(trn);
h1 = pboundary(model,struct(’line_style’,’r’));
h2 = pboundary(red_model,struct(’line_style’,’b’));
legend([h1(1) h2(1)],’Original SVM’,’Reduced SVM’);
% evaluate SVM classifiers
tst = load(’riply_tst’);
ypred = svmclass(tst.X,model);
err = cerror(ypred,tst.y);
red_ypred = svmclass(tst.X,model);
74
red_err = cerror(red_ypred,tst.y);
fprintf([’Original SVM: nsv = %d, err = %.4f\n’ ...
’Reduced SVM: nsv = %d, err = %.4f\n’], ...
model.nsv, err, red_model.nsv, red_err);
Original SVM: nsv = 94, err = 0.0960
Reduced SVM: nsv = 10, err = 0.0960
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Original SVM
Reduced SVM
Figure 5.7: Example shows the decision boundaries of the SVM classiﬁer trained by
SMO and the approximated SVM classiﬁer found by the reduced set method. The
original decision rule involves 94 support vectors while the reduced one only 10 support
vectors.
75
Chapter 6
Miscellaneous
6.1 Data sets
The STPRtool provides several datasets and functions to generate synthetic data. The
datasets stored in the Matlab data ﬁle are enlisted in Table 6.1. The functions which
generate synthetic data and scripts for converting external datasets to the Matlab are
given in Table 6.2.
Table 6.1: Datasets stored in /data/ directory of the STPRtool.
/riply trn.mat Training part of the Riply’s data set [24].
/riply tst.mat Testing part of the Riply’s data set [24].
/iris.mat Well known Fisher’s data set.
/anderson task/*.mat Input data for the demo on the Generalized Ander
son’s task demo anderson.
/binary separable/*.mat Input data for the demo on training linear classiﬁers
from ﬁnite vector sets demo linclass.
/gmm samles/*.mat Input data for the demo on the Expectation
Maximization algorithm for the Gaussian mixture
model demo emgmm.
/multi separable/*.mat Linearly separable multiclass data in 2D.
/ocr data/*.mat Examples of handwritten numerals. A description is
given in Section 7.1.
/svm samples/*.mat Input data for the demo on the Support Vector Ma
chines demo svm.
76
Table 6.2: Data generation.
createdata GUI which allows to create interactively labeled vec
tors sets and labeled sets of Gaussian distributions. It
operates in 2dimensional space only.
genlsdata Generates linearly separable binary data with pre
scribed margin.
gmmsamp Generates data from the Gaussian mixture model.
gsamp Generates data from the multivariate Gaussian distri
bution.
gencircledata Generates data on circle corrupted by the Gaussian
noise.
usps2mat Converts U.S. Postal Service (USPS) database of
handwritten numerals to the Matlab ﬁle.
6.2 Visualization
The STPRtool provides tools for visualization of common models and data. The list
of functions for visualization is given in Table 6.3. Each function contains an example
of using.
Table 6.3: Functions for visualization.
pandr Visualizes solution of the Generalized Anderson’s
task.
pboundary Plots decision boundary of given classiﬁer in 2D.
pgauss Visualizes set of bivariate Gaussians.
pgmm Visualizes the bivariate Gaussian mixture model.
pkernelproj Plots isolines of kernel projection.
plane3 Plots plane in 3D.
pline Plots line in 2D.
ppatterns Visualizes patterns as points in the feature space. It
works for 1D, 2D and 3D.
psvm Plots decision boundary of binary SVM classiﬁer in
2D.
showim Displays images entered as column vectors.
77
6.3 Bayesian classiﬁer
The object under study is assumed to be described by a vector of observations x ∈ A
and a hidden state y ∈ ¸. The x and y are realizations of random variables with joint
probability distribution P
XY
(x, y). A decision rule q: A → T takes a decision d ∈ T
based on the observation x ∈ A. Let W: T¸ →R be a loss function which penalizes
the decision q(x) ∈ T when the true hidden state is y ∈ ¸. Let A ⊆ R
n
and the sets
¸ and T be ﬁnite. The Bayesian risk R(q) is an expectation of the value of the loss
function W when the decision rule q is applied, i.e.,
R(q) =
_
X
y∈Y
P
XY
(x, y)W(q(x), y) dx . (6.1)
The optimal rule q
∗
which minimizes the Bayesian risk (6.1) is referred to as the
Bayesian rule
q
∗
(x) = argmin
y
y∈Y
P
XY
(x, y)W(q(x), y) , ∀x ∈ A .
The STPRtool implements the Bayesian rule for two particular cases:
Minimization of misclassiﬁcation: The set of decisions T coincides to the set of
hidden states ¸ = ¦1, . . . , c¦. The 0/1loss function
W
0/1
(q(x), y) =
_
0 for q(x) = y ,
1 for q(x) ,= y .
(6.2)
is used. The Bayesian risk (6.1) with the 0/1loss function corresponds to the
expectation of misclassiﬁcation. The rule q: A → ¸ which minimizes the expec
tation of misclassiﬁcation is deﬁned as
q(x) = argmax
y∈Y
P
Y X
(y[x) ,
= argmax
y∈Y
P
XY
(x[y)P
Y
(y) .
(6.3)
Classiﬁcation with rejectoption: The set of decisions T is assumed to be T =
¸ ∪ ¦dont know¦. The loss function is deﬁned as
W
ε
(q(x), y) =
⎧
⎨
⎩
0 for q(x) = y ,
1 for q(x) ,= y ,
ε for q(x) = dont know ,
(6.4)
where ε is penalty for the decision dont know. The rule q: A → ¸ which mini
mizes the Bayesian risk with the loss function (6.4) is deﬁned as
q(x) =
⎧
⎨
⎩
argmax
y∈Y
P
XY
(x[y)P
Y
(y) if 1 −max
y∈Y
P
Y X
(y[x) < ε ,
dont know if 1 −max
y∈Y
P
Y X
(y[x) ≥ ε .
(6.5)
78
To apply the optimal classiﬁcation rules one has to know the classconditional
distributions P
XY
and a priory distribution P
Y
(or their estimates). The datatype
used by the STPRtool is described in Table 6.4. The Bayesian classiﬁers (6.3) and (6.5)
are implemented in function bayescls.
Table 6.4: Datatype used to describe the Bayesian classiﬁer.
Bayesian classiﬁer (structure array):
.Pclass [cell 1 c] Class conditional distributions P
XY
. The distribution
P
XY
(x[y) is given by Pclassy which uses the data
type described in Table 4.2.
.Prior [1 c] Discrete distribution P
Y
(y), y ∈ ¸.
.fun = ’bayescls’ Identiﬁes function associated with this data type.
References: The Bayesian decision making with applications to PR is treated in
details in book [26].
Example: Bayesian classiﬁer
The example shows how to design the plugin Bayesian classiﬁer for the Riply’s
training set riply trn.mat. The classconditional distributions P
XY
are modeled by
the Gaussian mixture models (GMM) with two components. The GMM are estimated
by the EM algorithm emgmm. The a priori probabilities P
Y
(y), y ∈ ¦1, 2¦ are estimated
by the relative occurrences (it corresponds to the ML estimate). The designed Bayesian
classiﬁer is evaluated on the testing data riply tst.mat.
% load input training data
trn = load(’riply_trn’);
inx1 = find(trn.y==1);
inx2 = find(trn.y==2);
% Estimation of classconditional distributions by EM
bayes_model.Pclass{1} = emgmm(trn.X(:,inx1),struct(’ncomp’,2));
bayes_model.Pclass{2} = emgmm(trn.X(:,inx2),struct(’ncomp’,2));
% Estimation of priors
n1 = length(inx1); n2 = length(inx2);
bayes_model.Prior = [n1 n2]/(n1+n2);
% Evaluation on testing data
tst = load(’riply_tst’);
ypred = bayescls(tst.X,bayes_model);
cerror(ypred,tst.y)
79
ans =
0.0900
The following example shows how to visualize the decision boundary of the found
Bayesian classiﬁer minimizing the misclassiﬁcation (solid black line). The Bayesian
classiﬁer splits the feature space into two regions corresponding to the ﬁrst class and
the second class. The decision boundary of the rejectoptions rule (dashed line) is
displayed for comparison. The rejectoption rule splits the feature space into three
regions corresponding to the ﬁrst class, the second class and the region in between
corresponding to the dont know decision. The visualization is given in Figure 6.1.
% Visualization
figure; hold on; ppatterns(trn);
bayes_model.fun = ’bayescls’;
pboundary(bayes_model);
% Penalization for don’t know decision
reject_model = bayes_model;
reject_model.eps = 0.1;
% Vislualization of rejetoption rule
pboundary(reject_model,struct(’line_style’,’k’));
6.4 Quadratic classiﬁer
The quadratic classiﬁcation rule q: A ⊆ R
n
→ ¸ = ¦1, 2, . . . , c¦ is composed of a set
of discriminant functions
f
y
(x) = ¸x A
y
x) +¸b
y
x) + c
y
, ∀y ∈ ¸ ,
which are quadratic with respect to the input vector x ∈ R
n
. The quadratic discrimi
nant function f
y
is determined by a matrix A
y
[n n], a vector b
y
[n 1] and a scalar
c
y
[1 1]. The input vector x ∈ R
n
is assigned to the class y ∈ ¸ its corresponding
discriminant function f
y
attains maximal value
y = argmax
y
∈Y
f
y
(x) = argmax
y
∈Y
(¸x A
y
x) +¸b
y
x) + c
y
) . (6.6)
In the particular binary case ¸ = ¦1, 2¦, the quadratic classiﬁer is represented by
a single discriminant function
f(x) = ¸x Ax) +¸b x) + c ,
80
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 6.1: Figure shows the decision boundary of the Bayesian classiﬁer (solid line)
and the decision boundary of the rejectoption rule with ε = 0.1 (dashed line). The
classconditional probabilities are modeled by the Gaussian mixture models with two
components estimated by the EM algorithm.
given by a matrix A [n n], a vector b [n 1] and a scalar c [1 1]. The input vector
x ∈ R
n
is assigned to class y ∈ ¦1, 2¦ as follows
q(x) =
_
1 if f(x) = ¸x Ax) +¸b x) + c ≥ 0 ,
2 if f(x) = ¸x Ax) +¸b x) + c < 0 .
(6.7)
The STPRtool uses a speciﬁc structure array to describe the binary (see Table 6.5)
and the multiclass (see Table 6.6) quadratic classiﬁer. The implementation of the
quadratic classiﬁer itself provides function quadclass.
Table 6.5: Datatype used to describe binary quadratic classiﬁer.
Binary linear quadratic (structure array):
.A [n n] Parameter matrix A.
.b [n 1] Parameter vector b.
.c [1 1] Parameter scalar c.
.fun = ’quadclass’ Identiﬁes function associated with this data type.
Example: Quadratic classiﬁer
The example shows how to train the quadratic classiﬁer using the Fisher Linear Dis
criminant and the quadratic data mapping. The classiﬁer is trained from the Riply’s
81
Table 6.6: Datatype used to describe multiclass quadratic classiﬁer.
Multiclass quadratic classiﬁer (structure array):
.A [n n c] Parameter matrices A
y
, y ∈ ¸.
.b [n c] Parameter vectors b
y
, y ∈ ¸.
.c [1 c] Parameter scalars c
y
, y ∈ ¸.
.fun = ’quadclass’ Identiﬁes function associated with this data type.
training set riply trn.mat. The found quadratic classiﬁer is evaluated on the test
ing data riply trn.mat. The boundary of the quadratic classiﬁer is visualized (see
Figure 6.2).
trn = load(’riply_trn’); % load input data
quad_data = qmap(trn); % quadratic mapping
lin_model = fld(quad_data); % train FLD
% make quadratic classifier
quad_model = lin2quad(lin_model);
% visualization
figure;
ppatterns(trn); pboundary(quad_model);
% evaluation
tst = load(’riply_tst’);
ypred = quadclass(tst.X,quad_model);
cerror(ypred,tst.y)
ans =
0.0940
6.5 Knearest neighbors classiﬁer
Let T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ be a set of prototype vectors x
i
∈ A ⊆ R
n
and
corresponding hidden states y
i
∈ ¸ = ¦1, . . . , c¦. Let R
n
(x) = ¦x
: x −x
 ≤ r
2
¦ be
a ball centered in the vector x in which lie K prototype vectors x
i
, i ∈ ¦1, . . . , l¦, i.e.,
[¦x
i
: x
i
∈ R
n
(x)¦[ = K. The Knearest neighbor (KNN) classiﬁcation rule q: A → ¸
is deﬁned as
q(x) = argmax
y∈Y
v(x, y) , (6.8)
82
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 6.2: Figure shows the decision boundary of the quadratic classiﬁer.
where v(x, y) is number of prototype vectors x
i
with hidden state y
i
= y which lie in
the ball x
i
∈ R
n
(x). The classiﬁcation rule (6.8) is computationally demanding if the
set of prototype vectors is large. The datatype used by the STPRtool to represent the
Knearest neighbor classiﬁer is described in Table 6.7.
Table 6.7: Datatype used to describe the Knearest neighbor classiﬁer.
KNN classiﬁer (structure array):
.X [n l] Prototype vectors ¦x
1
, . . . , x
l
¦.
.y [1 l] Labels ¦y
1
, . . . , y
l
¦.
.fun = ’knnclass’ Identiﬁes function associated with this data type.
References: The Knearest neighbors classiﬁer is described in most books on PR, for
instance, in book [6].
Example: Knearest neighbor classiﬁer
The example shows how to create the (K = 8)NN classiﬁcation rule from the
Riply’s training data riply trn.mat. The classiﬁer is evaluated on the testing data
riply tst.mat. The decision boundary is visualized in Figure 6.3.
% load training data and setup 8NN rule
trn = load(’riply_trn’);
model = knnrule(trn,8);
83
% visualize decision boundary and training data
figure; ppatterns(trn); pboundary(model);
% evaluate classifier
tst = load(’riply_tst’);
ypred = knnclass(tst.X,model);
cerror(ypred,tst.y)
ans =
0.1160
−1.5 −1 −0.5 0 0.5 1
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Figure 6.3: Figure shows the decision boundary of the (K = 8)nearest neighbor
classiﬁer.
84
Chapter 7
Examples of applications
We intend to demonstrate how methods implemented in the STPRtool can be used to
solve a more complex applications. The applications selected are the optical charac
ter recognition (OCR) system based on the multiclass SVM (Section 7.1) and image
denoising system using the kernel PCA (Section 7.2).
7.1 Optical Character Recognition system
The use of the STPRtool is demonstrated on a simple Optical Character Recognition
(OCR) system for handwritten numerals from 0 to 9. The STPRtool provides GUI
which allows to use standard computer mouse as an input device to draw images of
numerals. The GUI is intended just for demonstration purposes and the scanner or
other device would be used in practice instead. The multiclass SVM are used as the
base classiﬁer of the numerals. A numeral to be classiﬁed is represented as a vector
x containing pixel values taken from the numeral image. The training stage of the
OCR system involves (i) collection of training examples of numerals and (ii) training
the multiclass SVM. To see the trained OCR system run the demo script demo ocr.
The GUI for drawing numerals is implemented in a function mpaper. The function
mpaper creates a form composed of 50 cells to which the numerals can be drawn (see
Figure 7.1a). The drawn numerals are normalized to size 16 16 pixels (diﬀerent
size can be speciﬁed). The function mpaper is designed to invoke a speciﬁed function
whenever the middle mouse button is pressed. In particular, the functions save chars
and ocr fun are called from the mpaper to process drawn numerals. The function
save chars saves drawn numerals in the stage of collecting training examples and the
function ocr fun is used to recognize the numerals if the trained OCR is applied.
The set of training examples has to be collected before the OCR is trained. The
function collect chars invokes the form for drawing numerals and allows to save the
numerals to a speciﬁed ﬁle. The STPRtool contains examples of numerals collected
85
from 6 diﬀerent persons. Each person drew 50 examples of each of numerals from
’0’ to ’9’. The numerals are stored in the directory /data/ocr numerals/. The ﬁle
name format name xx.mat (the character can be omitted from the name) is used.
The string name is name of the person who drew the particular numeral and xx stands
for a label assigned to the numeral. The labels from y = 1 to y = 9 corresponds to the
numerals from ’1’ to ’9’ and the label y = 10 is used for numeral ’0’. For example,
to collect examples of numeral ’1’ from person honza cech
collect_chars(’honza_cech1’)
was called. Figure 7.1a shows the form ﬁlled with the examples of the numeral ’1’.
The Figure 7.1b shows numerals after normalization which are saved to the speci
ﬁed ﬁle honza cech.mat. The images of normalized numerals are stored as vectors
to a matrix X of size [256 50]. The vector dimension n = 256 corresponds to
16 16 pixels of each image numeral and l = 50 is the maximal number of numerals
drawn to a single form. If the ﬁle name format name xx.mat is met then the function
mergesets(Directory,DataFile) can be used to merge all the ﬁle to a single one.
The resulting ﬁle contains a matrix X of all examples and a vector y containing labels
are assigned to examples according the number xx. The string Directory speciﬁes the
directory where the ﬁles name xx.mat are stored and the string FileName is a ﬁle name
to which the resulting training set is stored.
Having the labeled training set T
XY
= ¦(x
1
, y
1
), . . . , (x
l
, y
l
)¦ created the multiclass
SVM classiﬁer can be trained. The methods to train the multiclass SVM are described
in Chapter 5. In particular, the OneAgainstOne (OAO) decomposition was used as
it requires the shortest training time. However, if a speed of classiﬁcation is the major
objective the multiclass BSVM formulation is often a better option as it produce less
number of support vectors. The SVM has two free parameters which have to be tuned:
the regularization constant C and the kernel function k(x
a
, x
b
) with its argument.
The RBF kernel k(x
a
, x
b
) = exp(−0.5x
a
− x
b

2
/σ
2
) determined by the parameter
σ was used in the experiment. Therefore, the free SVM parameters are C and σ. A
common method to tune the free parameters is to use the crossvalidation to select the
best parameters from a preselected set Θ = ¦(C
1
, σ
1
), . . . , (C
k
, σ
k
)¦. The STPRtool
contains a function evalsvm which evaluates the set of SVM parameters Θ based on the
crossvalidation estimation of the classiﬁcation error. The function evalsvm returns the
SVM model its parameters (C
∗
, σ
∗
) turned out to be have the lowest crossvalidation
error over the given set Θ. The parameter tuning is implemented in the script tune ocr.
The parameters C = ∞ (data are separable) and σ = 5 were found to be the best.
After the parameters (C
∗
, σ
∗
) are tuned the SVM classiﬁer is trained using all
training data available. The training procedure for OCR classiﬁer is implemented in
the function train ocr. The directory /demo/ocr/models/ contains the SVM model
trained by OAO decomposition strategy (function oaosvm) and model trained based
on the multiclass BSVM formulation (function bsvm2).
86
The particular model used by OCR is speciﬁed in the function ocr fun which
implicitly uses the ﬁle ocrmodel.mat stored in the directory /demo/ocr/. To run the
resulting OCR type demo ocr or call
mpaper(struct(’fun’,’ocr_fun’))
If one intends to modify the classiﬁer or the way how the classiﬁcation results are
displayed then modify the function ocr fun. A snapshot of running OCR is shown in
Figure 7.2.
7.2 Image denoising
The aim is to train a model of a given class of images in order to restore images
corrupted by noise. The U.S. Postal Service (USPS) database of images of hand
written numerals [17] was used as the class of images to model. The additive Gaussian
noise was assumed as the source of image degradation. The kernel PCA was used to
model the image class. This method was proposed in [20] and further developed in [15].
The STPRtool contains a script which converts the USPS database to the Matlab
data ﬁle. It is implemented in the function usps2mat. The corruption of the USPS
images by the Gaussian noise is implemented in script make noisy data. The script
make noisy data produces the Matlab ﬁle usps noisy.mat its contains is summarized
in Table 7.1.
The aim is to train a model of images of all numerals at ones. Thus the unlabeled
training data set T
X
= ¦x
1
, . . . , x
l
¦ is used. The model should describe the nonlinear
subspace A
img
⊆ R
n
where the images live. It is assumed that the nonlinear subspace
A
img
forms a linear subspace T
img
of a new feature space T. A link between the
input space A and the feature space T is given by a mapping φ: A → T which is
implicitly determined by a selected kernel function k: A A →R. The kernel function
corresponds to the dot product in the feature space, i.e., k(x
a
, x
b
) = ¸φ(x
a
) φ(x
b
)).
The kernel PCA can be used to ﬁnd the basis vectors of the subspace T
img
from the
input training data T
X
. Having the kernel PCA model of the subspace T
img
it can be
used to map the images x ∈ A into the subspace A
img
. A noisy image ˆ x is assumed to
lie outside the subspace A
img
. The denoising procedure is based on mapping the noisy
image ˆ x into the subspace A
img
. The idea of using the kernel PCA for image denoising
is demonstrated in Figure 7.3. The example which produced the ﬁgure is implemented
in demo kpcadenois.
The subspace T
img
is modeled by the kernel PCA trained from the set T
X
mapped
into the feature space T by a function φ: A → T. Let T
Φ
= ¦φ(x
1
), . . . , φ(x
l
)¦ denote
the feature space representation of the training images T
X
. The kernel PCA computes
the lower dimensional representation T
Z
= ¦z
1
, . . . , z
l
¦, z
i
∈ R
m
of the mapped images
87
Table 7.1: The content of the ﬁle usps noisy.mat produced by the script
make noisy data.
Structure array trn containing the training data:
gnd X [256 7291] Training images of numerals represented as column
vectors. The input are 7291 grayscale images of size
16 16.
X [256 7291] Training images corrupted by the Gaussian noise with
signaltonoise ration SNR = 1 (it can be changed).
y [1 7291] Labels y ∈ ¸ = ¦1, . . . , 10¦ of training images.
Structure array tst containing the testing data:
gnd X [256 2007] Testing images of numerals represented as column vec
tors. The input are 2007 grayscale images of size
16 16.
X [256 2007] Training images corrupted by the Gaussian noise with
signaltonoise ration SNR = 1 (it can be changed).
y [1 2007] Labels y ∈ ¸ = ¦1, . . . , 10¦ of training images.
T
Φ
to minimize the mean square reconstruction error
ε
KMS
=
l
i=1
φ(x
i
) −
˜
φ(x
i
)
2
,
where
˜
φ(x) =
l
i=1
β
i
φ(x) , β = A(z −b) , z = A
T
k(x) +b . (7.1)
The vector k(x) = [k(x
1
, x), . . . , k(x
l
, x)]
T
contains kernel functions evaluated at the
training data T
X
. The matrix A [l m] and the vector b [m 1] are trained by the
kernel PCA algorithm. Given a projection
˜
φ(x) onto the kernel PCA subspace an
image x
∈ R
n
of the input space can be recovered by solving the preimage problem
x
= argmin
x
φ(x) −
˜
φ(x)
2
. (7.2)
The Radial Basis Function (RBF) kernel k(x
a
, x
b
) = exp(−0.5x
a
− x
b

2
/σ
2
) was
used here as the preimage problem (7.2) can be well solved for this case. The whole
procedure of reconstructing the corrupted image ˆ x ∈ R
n
involves (i) using (7.1) to
compute z which determines
˜
φ(x) and (ii) solving the preimage problem (7.2) which
yields the resulting x
. This procedure is implemented in function kpcarec.
In contrast to the linear PCA, the kernel PCA allows to model the nonlinearity
which is very likely in the case of images. However, the nonlinearity hidden in
88
the parameter σ of the used RBF kernel has to be tuned properly to prevent over
ﬁtting. Another parameter is the dimension m of the subspace kernel PCA subspace
to be trained. The best parameters (σ
∗
, m
∗
) was selected out of a prescribed set
Θ = ¦(σ
1
, m
1
), . . . , (σ
k
, m
k
)¦ such that the crossvalidation estimate of the mean square
error
ε
MS
(σ, m) =
_
X
P
X
(x)
_
x −x

2
_
dx ,
was minimal. The x stands for the ground truth image and x
denotes the image
restored from noisy one ˆ x using (7.1) and (7.2). The images are assumed to be gen
erated from distribution P
X
(x). Figure 7.4 shows the idea of tuning the parameters
of the kernel PCA. The linear PCA was used to solve the same task for comparison.
The same idea was used to tune the output dimension of the linear PCA. Having the
parameters tuned the resulting kernel PCA and linear PCA are trained on the whole
training set. The whole training procedure of the kernel PCA is implemented in the
script train kpca denois. The training for linear PCA is implemented in the script
train lin denois. The greedy kernel PCA Algorithm 3.4 is used for training the
model. It produces sparse model and can deal with large data sets. The results of
tuning the parameters are displayed in Figure 7.5. The ﬁgure was produced by a script
show denois tuning.
The function kpcarec is used to denoise images using the kernel PCA model. In
the case of the linear PCA, the function pcarec was used. An example of denoising of
the testing data of USPS is implemented in script show denoising. Figure 7.6 shows
examples of ground truth images, noisy images, images restored by the linear PCA and
kernel PCA.
89
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Control: left_button − draw, middle_button − classify, right_button − erase.
(a)
(b)
Figure 7.1: Figure (a) shows handwritten training examples of numeral ’1’ and Figure
(b) shows corresponding normalized images of size 16 16 pixels.
90
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Control: left_button − draw, middle_button − classify, right_button − erase.
(a)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
1 2 3 4 5 6 7 8 9 0
4 0
5 3 1 9
6 2 8
(b)
Figure 7.2: A snapshot of running OCR system. Figure (a) shows handwritten nu
merals and Figure (b) shows recognized numerals.
91
−4 −3 −2 −1 0 1 2 3 4 5 6 7
−4
−3
−2
−1
0
1
2
3
4
5
6
7
Ground truth
Noisy examples
Corrupted
Reconstructed
Figure 7.3: Demonstration of idea of image denoising by kernel PCA.
Data
generator
model
model
Distortion
KPCA Preimage
problem
θ ∈ Θ
Error
e
r
r
o
r
θ
assessment
Figure 7.4: Schema of tuning parameters of kernel PCA model used for image restora
tion.
92
0 10 20 30 40 50 60 70 80 90 100
10
11
12
13
14
15
16
17
18
dim
ε
M
S
3.5 4 4.5 5 5.5 6 6.5 7 7.5 8
6
8
10
12
14
16
18
20
σ
ε
M
S
dim = 50
dim = 100
dim = 200
dim = 300
(a) (b)
Figure 7.5: Figure (a) shows plot of ε
MS
with respect to output dimension of lin
ear PCA. Figure(b) shows plot of ε
MS
with respect to output dimension and kernel
argument of the kernel PCA.
Ground truth
Numerals with added Gaussian noise
Linear PCA reconstruction
Kernel PCA reconstruction
Figure 7.6: The results of denosing the handwritten images of USPS database.
93
Chapter 8
Conclusions and future planes
The STPRtool as well as its documentation is being updating. Therefore we are grateful
for any comment, suggestion or other feedback which would help in the development
of the STPRtool.
There are many relevant methods and proposals which have not been implemented
yet. We welcome anyone who likes to contribute to the STPRtool in any way. Please
send us an email and your potential contribution.
The future planes involve the extensions of the toolbox by the following methods:
• Feature selection methods.
• AdaBoost and related learning methods.
• Eﬃcient optimizers to for large scale Support Vector Machines learning problems.
• Model selection methods for SVM.
• Probabilistic Principal Component Analysis (PPCA) and mixture of PPCA esti
mated by the ExpectationMaximization (EM) algorithm.
• Discrete probability models estimated by the EM algorithm.
94
Bibliography
[1] T.W. Anderson and R.R. Bahadur. Classiﬁcation into two multivariate normal
distributions with diﬀerrentia covariance matrices. Anals of Mathematical Statis
tics, 33:420–431, June 1962.
[2] G. Baudat and F. Anouar. Generalized discriminant analysis using a kernel ap
proach. Neural Computation, 12(10):2385–2404, 2000.
[3] C.M. Bishop. Neural Networks for Pattern Recognition. Clarendon Press, Oxford,
Great Britain, 3th edition, 1997.
[4] N. Cristianini and J. ShaweTaylor. Support Vector Machines. Cambridge Uni
versity Press, 2000.
[5] A.P. Dempster, N.M. Laird, and D.B. Rubin. Maximum likelihood from incom
plete data via the EM Algorithm. Journal of the Royal Statistical Society, 39:185–
197, 1977.
[6] R.O. Duda, P.E. Hart, and D.G. Stork. Pattern Classiﬁcation. John Wiley &
Sons, 2nd. edition, 2001.
[7] R.P.W. Duin. Prtools: A matlab toolbox for pattern recognition, 2000.
[8] V. Franc. Programov´e n´astroje pro rozpozn´ av´ an´ı (Pattern Recognition Program
ming Tools, In Czech). Master’s thesis,
ˇ
Cesk´e vysok´e uˇcen´ı technick´e, Fakulta
elektrotechnick´a, Katedra kybernetiky, February 2000.
[9] V. Franc and V. Hlav´aˇc. Multiclass support vector machine. In R. Kasturi,
D. Laurendeau, and Suen C., editors, 16th International Conference on Pattern
Recognition, volume 2, pages 236–239. IEEE Computer Society, 2002.
[10] V. Franc and V. Hlav´ aˇc. An iterative algorithm learning the maximal margin
classiﬁer. Pattern Recognition, 36(9):1985–1996, 2003.
95
[11] V. Franc and V. Hlav´ aˇc. Greedy algorithm for a training set reduction in the
kernel methods. In N. Petkov and M.A. Westenberg, editors, Computer Analysis
of Images and Patterns, pages 426–433, Berlin, Germany, 2003. Springer.
[12] C.W. Hsu and C.J. Lin. A comparison of methods for multiclass support vector
machins. IEEE Transactions on Neural Networks, 13(2), March 2002.
[13] I.T. Jollife. Principal Component Analysis. SpringerVerlag, New York, 1986.
[14] S.S. Keerthi, S.K. Shevade, C. Bhattacharya, and K.R.K. Murthy. A fast itera
tive nearest point algorithm for support vector machine classiﬁer design. IEEE
Transactions on Neural Networks, 11(1):124–136, January 2000.
[15] Kim Kwang, In, Franz Matthias, O., and Sch¨olkopf Bernhard. Kernel hebbian
algorithm for singlefram superresolution. In Leonardis Aleˇs and Bischof Horst,
editors, Statisical Learning in Computer Vision, ECCV Workshop. Springer, May
2004.
[16] J.T. Kwok and I.W. Tsang. The preimage problem in kernel methods. In Pro
ceedings of the Twentieth International Conference on Machine Learning (ICML
2003), pages 408–415, Washington, D.C., USA, August 2003.
[17] Y. LeCun, B. Boser, J.S. Denker, D. Henderson, R.E. Howard, W. Hubbard, and
L.J Jackel. Backpropagation applied to handwritten zip code recognition. Neural
Computation, 1:541–551, 1989.
[18] G. McLachlan and T. Krishnan. The EM Algorithm and Extensions. John Wiley
& Sons, New York, 1997.
[19] S. Mika, G. R¨ atsch, J. Weston, B. Sch¨olkpf, and K. M¨ uller. Fisher discriminant
analysis with kernel. In Y.H. Hu, J. Larsen, and S. Wilson, E. Douglas, editors,
Neural Networks for Signal Processing, pages 41–48. IEEE, 1999.
[20] S. Mika, B. Sch¨ olkopf, A. Smola, K.R. M¨ uller, M. Scholz, and G. R¨ atsch. Kernel
pca and denoising in feature spaces. In M.S. Kearns, S.A. Solla, and D.A. Cohn,
editors, Advances in Neural Information Processing Systems 11, pages 536 – 542,
Cambridge, MA, 1999. MIT Press.
[21] I.T. Nabney. NETLAB : algorithms for pattern recognition. Advances in pattern
recognition. Springer, London, 2002.
[22] J. Platt. Probabilities for sv machines. In A.J. Smola, P.J. Bartlett, B. Scholkopf,
and D. Schuurmans, editors, Advances in Large Margin Classiﬁers (Neural Infor
mation Processing Series). MIT Press, 2000.
96
[23] J.C. Platt. Sequential minimal optimizer: A fast algorithm for training support
vector machines. Technical Report MSRTR9814, Microsoft Research, Redmond,
1998.
[24] B.D. Riply. Neural networks and related methods for classiﬁcation (with discu
sion). J. Royal Statistical Soc. Series B, 56:409–456, 1994.
[25] M.I. Schlesinger. A connection between learning and selflearning in the pattern
recognition (in Russian). Kibernetika, 2:81–88, 1968.
[26] M.I. Schlesinger and V. Hlav´ aˇc. Ten lectures on statistical and structural pattern
recognition. Kluwer Academic Publishers, 2002.
[27] B. Sch¨olkopf, C. Burges, and A.J. Smola. Advances in Kernel Methods – Support
Vector Learning. MIT Press, Cambridge, 1999.
[28] B. Sch¨olkopf, P. Knirsch, and C. Smola, A. Burges. Fast approximation of support
vector kernel expansions, and an interpretation of clustering as approximation in
feature spaces. In P. Levi, M. Schanz, R.J. Ahler, and F. May, editors, Mus
tererkennung 199820. DAGM., pages 124–132, Berlin, Germany, 1998. Springer
Verlag.
[29] B. Scholkopf, A. Smola, and K.R. Muller. Nonlinear component analysis as a
kernel eigenvalue problem. Neural Computation, 10:1299–1319, 1998.
[30] B. Sch¨olkopf and A.J. Smola. Learning with Kernels. The MIT Press, MA, 2002.
[31] V. Vapnik. The nature of statistical learning theory. Springer Verlag, 1995.
97
Statistical Pattern Recognition Toolbox for Matlab
Vojtˇch Franc and V´clav Hlav´ˇ e a ac June 24, 2004
Contents
1 Introduction 1.1 What is the Statistical Pattern Recognition Toolbox and how it has been developed? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 The STPRtool purpose and its potential users . . . . . . . . . . . . . . 1.3 A digest of the implemented methods . . . . . . . . . . . . . . . . . . . 1.4 Relevant Matlab toolboxes of others . . . . . . . . . . . . . . . . . . . . 1.5 STPRtool document organization . . . . . . . . . . . . . . . . . . . . . 1.6 How to contribute to STPRtool? . . . . . . . . . . . . . . . . . . . . . 1.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Linear Discriminant Function 2.1 Linear classiﬁer . . . . . . . . . . . . . . . . . . 2.2 Linear separation of ﬁnite sets of vectors . . . . 2.2.1 Perceptron algorithm . . . . . . . . . . . 2.2.2 Kozinec’s algorithm . . . . . . . . . . . . 2.2.3 Multiclass linear classiﬁer . . . . . . . . 2.3 Fisher Linear Discriminant . . . . . . . . . . . . 2.4 Generalized Anderson’s task . . . . . . . . . . . 2.4.1 Original Anderson’s algorithm . . . . . . 2.4.2 General algorithm framework . . . . . . 2.4.3 εsolution using the Kozinec’s algorithm 2.4.4 Generalized gradient optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 3 3 4 5 5 5 6 8 9 10 11 12 14 15 17 19 21 21 22 25 26 29 31 32 33 34
3 Feature extraction 3.1 Principal Component Analysis . . . . . . . . . . . 3.2 Linear Discriminant Analysis . . . . . . . . . . . 3.3 Kernel Principal Component Analysis . . . . . . . 3.4 Greedy kernel PCA algorithm . . . . . . . . . . . 3.5 Generalized Discriminant Analysis . . . . . . . . . 3.6 Combining feature extraction and linear classiﬁer
1
. . . . . 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . .2 Image denoising . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . . . 4. . . . . . . . . . .3 Minimax estimation . . . . . .1 Complete data . . . . . . . . . . . . . . 6. . . . . . . . . . . . . . . . . . .2 MaximumLikelihood estimation of GMM . . 4. . . . . . . . . . . . .1 Support vector classiﬁers . . . . . . . . . . . .3 Bayesian classiﬁer . . . . . . .1 Data sets . . . . . . . . . . . . . . . . . . . . 4. . . . . . . . . . . . . . 6. . . . . . . . 5. . . . . . . . . . . . . . . . .4 Density estimation and clustering 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Probabilistic output of classiﬁer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 OneAgainstAll decomposition for SVM . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Gaussian distribution and Gaussian mixture model 4. . . . . . . . . . . 5. . . 5. . . 6 Miscellaneous 6. .4 OneAgainstOne decomposition for SVM . . . . .4 Quadratic classiﬁer . . . . . . . . . . . . . . . . . . . . . . . . 6. . . . . . 5.5 Multiclass BSVM formulation . .5 Kmeans clustering . . . . .7 Preimage problem . . . . . . .1 Optical Character Recognition system . . 7. 8 Conclusions and future planes 2 . . . . . . . . . . . . . . . . . . . . . . . 6. . . . 5 Support vector and other kernel machines 5. . . . . . . . . . . . . . . . . . . . . . . . . .6 Kernel Fisher Discriminant . . . . .2 Visualization . . . . .2 Binary Support Vector Machines . . . . . . . . .2 Incomplete data . . . . . . . . 41 41 43 44 45 47 48 50 57 57 58 63 64 66 68 70 73 76 76 77 78 80 82 85 85 87 94 7 Examples of applications 7. . . . . . . . . . . . . . . . . . . 4. . . . . .5 Knearest neighbors classiﬁer . . . . . . . 5. . . . . . . . . . . . . . . . . .8 Reduced set method . . 5. . . . . .2. . . . . . . . . . .
Chapter 1 Introduction
1.1 What is the Statistical Pattern Recognition Toolbox and how it has been developed?
The Statistical Pattern Recognition Toolbox (abbreviated STPRtool) is a collection of pattern recognition (PR) methods implemented in Matlab. The core of the STPRtool is comprised of statistical PR algorithms described in the monograph Schlesinger, M.I., Hlav´ˇ, V: Ten Lectures on Statistical and Structural ac Pattern Recognition, Kluwer Academic Publishers, 2002 [26]. The STPRtool has been e developed since 1999. The ﬁrst version was a result of a diploma project of Vojtˇch Franc [8]. Its author became a PhD student in CMP and has been extending the STPRtool since. Currently, the STPRtool contains much wider collection of algorithms. The STPRtool is available at http://cmp.felk.cvut.cz/cmp/cmp software.html. We tried to create concise and complete documentation of the STPRtool. This document and associated help ﬁles in .html should give all information and parameters the user needs when she/he likes to call appropriate method (typically a Matlab mﬁle). The intention was not to write another textbook. If the user does not know the method then she/he will likely have to read the description of the method in a referenced original paper or in a textbook.
1.2
The STPRtool purpose and its potential users
• The principal purpose of STPRtool is to provide basic statistical pattern recognition methods to (a) researchers, (b) teachers, and (c) students. The STPRtool is likely to help both in research and teaching. It can be found useful by practitioners who design PR applications too.
3
• The STPRtool oﬀers a collection of the basic and the stateofthe art statistical PR methods. • The STPRtool contains demonstration programs, working examples and visualization tools which help to understand the PR methods. These teaching aids are intended for students of the PR courses and their teachers who can easily create examples and incorporate them into their lectures. • The STPRtool is also intended to serve a designer of a new PR method by providing tools for evaluation and comparison of PR methods. Many standard and stateofthe art methods have been implemented in STPRtool. • The STPRtool is open to contributions by others. If you develop a method which you consider useful for the community then contact us. We are willing to incorporate the method into the toolbox. We intend to moderate this process to keep the toolbox consistent.
1.3
A digest of the implemented methods
This section should give the reader a quick overview of the methods implemented in STPRtool. • Analysis of linear discriminant function: Perceptron algorithm and multiclass modiﬁcation. Kozinec’s algorithm. Fisher Linear Discriminant. A collection of known algorithms solving the Generalized Anderson’s Task. • Feature extraction: Linear Discriminant Analysis. Principal Component Analysis (PCA). Kernel PCA. Greedy Kernel PCA. Generalized Discriminant Analysis. • Probability distribution estimation and clustering: Gaussian Mixture Models. ExpectationMaximization algorithm. Minimax probability estimation. Kmeans clustering. • Support Vector and other Kernel Machines: Sequential Minimal Optimizer (SMO). Matlab Optimization toolbox based algorithms. Interface to the SV M light software. Decomposition approaches to train the Multiclass SVM classiﬁers. Multiclass BSVM formulation trained by Kozinec’s algorithm, MitchellDemyanovMolozenov algorithm and Nearest Point Algorithm. Kernel Fisher Discriminant.
4
1.4
Relevant Matlab toolboxes of others
Some researchers in statistical PR provide code of their methods. The Internet oﬀers many software resources for Matlab. However, there are not many compact and well documented packages comprised of PR methods. We found two useful packages which are brieﬂy introduced below. NETLAB is focused to the Neural Networks applied to PR problems. The classical PR methods are included as well. This Matlab toolbox is closely related to the popular PR textbook [3]. The documentation published in book [21] contains detailed description of the implemented methods and many working examples. The toolbox includes: Probabilistic Principal Component Analysis, Generative Topographic Mapping, methods based on the Bayesian inference, Gaussian processes, Radial Basis Functions (RBF) networks and many others. The toolbox can be downloaded from http://www.ncrg.aston.ac.uk/netlab/. PRTools is a Matlab Toolbox for Pattern Recognition [7]. Its implementation is based on the object oriented programming principles supported by the Matlab language. The toolbox contains a wide range of PR methods including: analysis of linear and nonlinear classiﬁers, feature extraction, feature selection, methods for combining classiﬁers and methods for clustering. The toolbox can be downloaded from http://www.ph.tn.tudelft.nl/prtools/.
1.5
STPRtool document organization
The document is divided into chapters describing the implemented methods. The last chapter is devoted to more complex examples. Most chapters begin with deﬁnition of the models to be analyzed (e.g., linear classiﬁer, probability densities, etc.) and the list of the implemented methods. Individual sections have the following ﬁxed structure: (i) deﬁnition of the problem to be solved, (ii) brief description and list of the implemented methods, (iii) reference to the literature where the problem is treaded in details and (iv) an example demonstrating how to solve the problem using the STPRtool.
1.6
How to contribute to STPRtool?
If you have an implementation which is consistent with the STPRtool spirit and you want to put it into public domain through the STPRtool channel then send an email to xfrancv@cmp.felk.cvut.cz. 5
7 Acknowledgements First. our visitors. thanks go to several Czech.Please prepare your code according to the pattern you ﬁnd in STPRtool implementations. we like to thank for discussions to our colleagues in the Center for Machine Perception. M. Schlesinger from the Ukrainian Academy of Sciences. Forth. the STPRtool was rewritten and its documentation radically improved in May and June 2004 with the ﬁnancial help of ECVision European Research Network on Cognitive Computer Vision Systems (IST200135454) lead by D. other STPRtool users and students who used the toolbox in their laboratory exercises. Kiev. Ukraine who showed us the beauty of pattern recognition and gave us many valuable advices. your method will be bound to your name in STPRtool. Vernon. Third. we like to thank to Prof. Second. 1. 6 . EU and industrial grants which indirectly contributed to STPRtool development. Provide piece of documentation too which will allow us to incorporate it to STPRtool easily. the Czech Technical University in Prague.I. Of course.
. Logical or. . . Discriminant function of the binary classiﬁer. Scatter matrix. The concatenation of column vectors w ∈ Rn . for instance x. j) Dot product between vectors x and x . yl )}. . Set of c hidden states – class labels Y = {1. Mean vector (mathematical expectation). y1 ). ndimensional input space (space of observations). . .Notation Uppercase bold letters denote matrices. Kronecker delta δ(i. Euclidean norm. c}. Unlabeled training set TX = {x1 . Discriminant function associated with the class y ∈ Y. j) = 1 for i = j and δ(i. . Covariance matrix. for instance A. z ∈ Rm is denoted as v = [w. Vectors are implicitly column vectors. . Kernel function (Mercer kernel). Matrix determinant. Vectors are denoted by lowercase bold italic letters. 7 . (xl . Mapping from input space X to the feature space F . . j) = 0 otherwise. Labeled training set (more precisely multiset) TXY = {(x1 . . Classiﬁcation rule. . x·x Σ µ S X ⊆ Rn Y F φ: X → F k: X × X → R TXY TX q: X → Y fy : X → R f: X → R · · det(()·) δ(i. z] where the column vector v is (n + m)dimensional. Cardinality of a set. Feature space. . xl }.
. . the linear classiﬁer is represented by a single discriminant function f (x) = w · x + b . . Following sections describe supervised learning methods which determine the parameters of the linear classiﬁer based on available knowledge. .2) The datatype used to describe the linear classiﬁer and the implementation of the classiﬁer is described in Section 2. The input vector x ∈ Rn is assigned to class y ∈ {1. The scalars by . 2 if f (x) = w · x + b < 0 . yl )} containing pairs of observations xi ∈ Rn and corresponding class labels yi ∈ Y. The implemented learning methods can be distinguished according to the type of the training data: I. 8 . ∀y ∈ Y . The input vector x ∈ Rn is assigned to the class y ∈ Y its corresponding discriminant function fy attains maximal value y = argmax fy (x) = argmax ( wy · x + by ) . c} is composed of a set of discriminant functions fy (x) = w y · x + by . which are linear with respect to both the input vector x ∈ Rn and their parameter vectors w ∈ Rn . . 2.Chapter 2 Linear Discriminant Function The linear classiﬁcation rule q: X ⊆ Rn → Y = {1. The problem is described by a ﬁnite set T = {(x1 . (2.2 and Section 2. 2} as follows q(x) = 1 if f (x) = w · x + b ≥ 0 . y1 ). given by the parameter vector w ∈ Rn and bias b ∈ R. . . y ∈Y y ∈Y (2. ∀y ∈ Y introduce bias to the discriminant functions.1) In the particular binary case Y = {1.1. (xl .3. The methods using this type of training data is described in Section 2. . 2}.
eanders Epsilonsolution of the Generalized Anderson’s task. The resulting linear classiﬁer is then evaluated on the testing data riply tst. The Riply’s data set riply trn.1 Linear classiﬁer The STPRtool uses a speciﬁc structure array to describe the binary (see Table 2.mat is used for training. 9 .II.1: Implemented methods: Linear discriminant function linclass Linear classiﬁer (linear machine). The Anderson’s task and its generalization are representatives of such methods (see Section 2.3) linear classiﬁer. androrig Original method to solve the AndersonBahadur’s task. Example: Linear classiﬁer The example shows training of the Fisher Linear Discriminant which is the classical example of the linear classiﬁer. 6]. The summary of implemented methods is given in Table 2. mperceptr Perceptron algorithm to train multiclass linear machine. fldqp Fisher Linear Discriminant using Quadratic Programming. The parameters of class conditional distributions are (completely or partially) known. ggradander Gradient method to solve the Generalized Anderson’s task. fld Fisher Linear Discriminant. Table 2. perceptron Perceptron algorithm to train binary linear classiﬁer. ekozinec Kozinec’s algorithm for epsoptimal separating hyperplane. demo linclass Demo on the algorithms learning linear classiﬁers. demo anderson Demo on Generalized Anderson’s task. andrerr Classiﬁcation error of the Generalized Anderson’s task.2) and the multiclass (see Table 2.mat.4). ganders Solves the Generalized Anderson’s task. The implementation of the linear classiﬁer itself provides function linclass.1. References: The linear discriminant function is treated throughout for instance in books [26. 2.
Binary linear classiﬁer (structure array): . (ii) training εoptimal hyperplane and (iii) training the multiclass linear classiﬁer.3) with respect to the vector w ∈ Rn and the scalar b ∈ R. .2. yl )} consists of pairs of observations xi ∈ Rn and corresponding class labels yi ∈ Y = {1. (2. Table 2.3) have a solution then the training data T is linearly separable. cerror( ypred. .2) with zero training error is equivalent to ﬁnding the hyperplane H = {x: w · x + b = 0} which separates the training vectors of the ﬁrst y = 1 and the second y = 2 class. . c of the linear discriminant functions fy (x) = w y · x + b. The deﬁnitions of these tasks are given below. yi = 1 .2: Datatype used to describe binary linear classiﬁer.fun = ’linclass’ Identiﬁes function associated with this data type. .y ) ans = 0. . .1080 % % % % % load training data load testing data train FLD classifier classify testing data evaluate testing error 2.X.b [1 × 1] Bias b ∈ R of the hyperplane. ypred = linclass( tst.fun = ’linclass’ Identiﬁes function associated with this data type. . The implemented methods solve the task of (i) training separating hyperplane for binary case c = 2.W [n × 1] The normal vector w ∈ Rn of the separating hyperplane f (x) = w · x + b. c. . .1) 10 . . tst = load(’riply_tst’). yi = 2 . (xl .3: Datatype used to describe multiclass linear classiﬁer. y = 1. Multiclass linear classiﬁer (structure array): . model ). trn = load(’riply_trn’). . w · xi + b < 0 . model = fld( trn ). . The problem is formally deﬁned as solving the set of linear inequalities w · xi + b ≥ 0 .2 Linear separation of ﬁnite sets of vectors The input training data T = {(x1 . . y1 ).b [c × 1] Parameters by ∈ R. c}. tst. y = 1. . . . The Perceptron (Section 2. .Table 2. If the inequalities (2.W [n × c] Matrix of parameter vectors w y ∈ Rn . . . . The problem of training the binary (c = 2) linear classiﬁer (2.
The problem of training the separating hyperplane (2. l . (2. where the vector v ∈ Rn+1 is constructed as v = [w.and Kozinec’s (Section 2.6) with respect to the vectors wy ∈ Rn . An interactive demo on the algorithms separating the ﬁnite sets of vectors by a hyperplane is implemented in demo linclass.2). The εoptimal solution can be found by Kozinec’s algorithm (Section 2. . . .2.b = argmax min min i∈I1 w · xi + b w · xi + b . b∗ ) − m(w. If the training data is linearly separable then the optimal separating hyperplane can be deﬁned the solution of the following task (w∗ . . The optimal separating hyperplane cannot be found exactly except special cases.6) for linearly separable training data T can be solved by the modiﬁed Perceptron (Section 2. b∗ ). . 11 (2. The task (2. y ∈ Y and scalars by ∈ R. .2. b] . b) w. 2} training vectors xi ∈ Rn . (2. (2.4) where I1 = {i: yi = 1} and I2 = {i: yi = 2} are sets of indices. 2.7) . yi = y .2) algorithm are useful to train the binary linear classiﬁers from the linearly separable data. .1 Perceptron algorithm The input is a set T = {(x1 . . i = 1.2. The problem of training the multiclass linear classiﬁer c > 2 with zero training error is formally stated as the problem of solving the set of linear inequalities w yi · xi + byi > wy · xi + by .3) can be formally rewritten to a simpler form v · zi > 0 . yl )} of binary labeled yi ∈ {1.2. b) ≤ ε . The optimal separating hyperplane H∗ = {x ∈ Rn : w∗ · x + b∗ = 0} separates training data T with maximal margin m(w ∗ . This task is equivalent to training the linear Support Vector Machines (see Section 5). l . . (xl . .b w. Therefore the numerical algorithms seeking the approximate solution are applied instead. y ∈ Y.8) i = 1. . b∗ ) = argmax m(w. min − w w i∈I2 . The parameter ε ≥ 0 deﬁnes closeness to the optimal solution in terms of the margin. The εoptimal solution is deﬁned as the vector w and scalar b such that the inequality m(w ∗ . (2.5) holds.3) or Kozinec’s algorithm. y1). .
2} training (0) (1) (t) vectors xi ∈ Rn . b) of the linear classiﬁer are obtained from the found vector v by inverting the transformation (2. The generated training data contain 50 labeled vectors which are linearly separable with margin 1. . . . model = perceptron( data ). .2. −[xi . z l } are deﬁned as zi = [xi . .9) The problem of solving (2. The 1 2 vectors w∗ and w ∗ are the solution of the following task 1 2 w∗. data = genlsdata( 2.7) with respect to the unknown vector v ∈ Rn+1 is equivalent to the original task (2.w2 ∈X2 w1 − w2 . . . . . References: Analysis of the Perceptron algorithm including the Novikoﬀ’s proof of convergence can be found in Chapter 5 of the book [26]. 1] if yi = 2 . . .1. figure. .7) is satisﬁed. yl )} of binary labeled yi ∈ {1. (2. 1] if yi = 1 . y1). w 2 . . The Novikoﬀ theorem ensures that the Perceptron algorithm stops after ﬁnite number of iterations t if the training data are linearly separable. . The parameters (w. pline( model ).8). (xl . . 1). . where X1 stands for the convex hull of the training vectors of the ﬁrst class X1 = {xi : yi = 1} and X2 for the convex hull of the second class likewise. See Figure 2. v (1) . . The Perceptron algorithm is an iterative procedure which builds a series of vectors (0) v .2 Kozinec’s algorithm The input is a set T = {(x1 . w1 . w2 which converge to the vector w ∗ and w ∗ respectively. 50. . ppatterns( data ). % generate training data % call perceptron % plot training data % plot separating hyperplane 2. The Kozinec’s algorithm builds a series of vectors w 1 . w∗ = 1 2 argmin w1 ∈X1 .3). w 1 (0) (1) (t) and w 2 . The found classiﬁer is visualized as a separating hyperplane (line in this case) H = {x ∈ R2 : w · x + b = 0}. v (t) until the set of inequalities (2. The Perceptron from the neural network perspective is described in the book [3]. Example: Training binary linear classiﬁer with Perceptron The example shows the application of the Perceptron algorithm to ﬁnd the binary linear classiﬁer for synthetically generated 2dimensional training data. The vector w∗ = 12 . . Perceptron algorithm is implemented in function perceptron.and transformed training data Z = {z 1 . . The initial vector v (0) can be set arbitrarily (usually v = 0).
The separating hyperplane (2. Therefore the following two stopping conditions are implemented: I. References: Analysis of the Kozinec’s algorithm including the proof of convergence can be found in Chapter 5 of the book [26]. The Kozinec’s algorithm is proven to converge to the vectors w∗ and w ∗ in inﬁnite 1 2 number of iterations t = ∞.3) is sought for (ε < 0).5) is sought for (ε ≥ 0). Notice that setting ε = 0 force the algorithm to seek the optimal hyperplane which is generally ensured to be found in an inﬁnite number of iterations t = ∞. The Kozinec’s algorithm is proven to converge in a ﬁnite number of iterations if the separating hyperplane exists. The εoptimal hyperplane (2. The Kozinec’s algorithm which trains the binary linear classiﬁer is implemented in the function ekozinec.−30 −35 −40 −45 −50 −55 −55 −50 −45 −40 −35 −30 Figure 2. The Kozinec’s algorithm can be also used to solve a simpler problem of ﬁnding the separating hyperplane (2.1: Linear classiﬁer trained by the Perceptron algorithm w∗ − w ∗ and the bias b∗ = 1 ( w ∗ 2 − w∗ 2 ) determine the optimal hyperplane (2.01)optimal hyperplane for synthetically generated 2dimensional training data. II. Example: Training εoptimal hyperplane by the Kozinec’s algorithm The example shows application of the Kozinec’s algorithm to ﬁnd the (ε = 0. If the εoptimal optimality stopping condition is used then the Kozinec’s algorithm converges in ﬁnite number of iterations.3). The generated training data contains 50 labeled vectors which are linearly separable with 13 .4) 1 2 2 1 2 separating the training data with the maximal margin.
. % plot data pline(model). w c . . The inequalities (2.margin which satisﬁes the εoptimality condition. c}. w 2 .50. It could be veriﬁed that the found hyperplane has margin model. l . yl )} consists of pairs of observations xi ∈ Rn and corresponding class labels yi ∈ Y = {1.1) which classiﬁes the training data T without error. 14 . figure. .1). y1 ).01)).2) as a separating hyperplane H = {x ∈ R2 : w · x + b = 0} (straight line in this case). . The found classiﬁer is visualized (Figure 2.10) where the vector v ∈ R(n+1)c contains the originally optimized parameters v = [w1 .6). data = genlsdata(2.0. struct(’eps’. % generate data % run ekozinec model = ekozinec(data. . (xl .2. % plot found hyperplane 8 6 4 2 0 −2 −4 −6 −8 −10 −10 −8 −6 −4 −2 0 2 4 6 8 10 Figure 2. . The task is to design the linear classiﬁer (2. i i = 1. bc ] .margin 1. . . .1) is equivalent to solving the set of linear inequalities (2. 2. . b2 . . .6) can be formally transformed to a simpler set v · zy > 0 . . y = Y \ {yi } . b1 . ppatterns(data). . .3 Multiclass linear classiﬁer The input training data T = {(x1 . .2: εoptimal hyperplane trained with the Kozinec’s algorithm. (2. The problem of training (2.
2} . for j = y . y ∈ {1. (xl . otherwise. i The described transformation is known as the Kesler’s construction. % load training data % run training algorithm % plot training data % plot decision boundary 2. w · SW w 1 xi .3 Fisher Linear Discriminant The input is a set T = {(x1 . y ∈ {1.10) can be solved by the Perceptron algorithm. 1] . The class separability in a direction w ∈ Rn is deﬁned as F (w) = w · SB w . yl )} of binary labeled yi ∈ {1. The transformed task (2. Iy  i∈I y (2. data = load(’pentagon’). .11) where z y (j) stands for the jth slot between coordinates (n + 1)(j − 1) + 1 and (n + 1)j. ppatterns(data). References: The Kesler’s construction is described in books [26. z i (j) = ⎩ 0. . y −[xi . pboundary(model). . . The found classiﬁer is visualized as its separating surface which is known to be piecewise linear and the class regions should be convex (see Figure 2. (2. µy = 15 .10) (number 10 stands for number of classes). ∈ Y \ {yi } is created from the training for j = yi . 2} training vectors xi ∈ Rn . . Example: Training multiclass linear classiﬁer by the Perceptron The example shows application of the Perceptron rule to train the multiclass linear classiﬁer for the synthetically generated 2dimensional training data pentagon. y1). . l. model = mperceptron(data). A user’s own data can be generated interactively using function createdata(’finite’. 6].12) where SB is the betweenclass scatter matrix SB = (µ1 − µ2 )(µ1 − µ2 )T . .The set of vectors z y ∈ R(n+1)c . y i set T such that ⎧ ⎨ [xi .3). i = 1. . Let Iy = {i: yi = y}. 2} be sets of indices of training vectors belonging to the ﬁrst y = 1 and the second y = 2 class. figure. respectively. 1] .mat. The described method is implemented in function mperceptron. The simple form of the Perceptron updating rule allows to implement the transformation implicitly without mapping the training data T into (n + 1)cdimensional space.
3: Multiclass linear classiﬁer trained by the Perceptron algorithm. The classical solution of the problem (2.12) w = argmax F (w ) = argmax w w w · SB w .13) The bias b of the linear rule must be determined based on another principle.4 −0.2 0 0. y ∈ {1. The STPRtool contains two implementations of FLD: I. and SW is the within class scatter matrix deﬁned as SW = S1 + S2 .4 0. computes such b that the equality w · µ1 + b = −( w · µ2 + b) . 16 . the parameter vector w of the linear discriminant function f (x) = w · x + b is determined to maximize the class separability criterion (2. 2} .6 0.6 −0.8 −0.2 0.8 0. w · SW w (2.6 −0. holds.1 0.4 0.4 −0.8 −1 −1 −0.2 −0. This case is implemented in the function fld. The STPRtool implementation.13) using the matrix inversion w = SW −1 (µ1 − µ2 ) .8 1 Figure 2.6 0.2 0 −0. In the case of the Fisher Linear Discriminant (FLD). for instance. Sy = i∈Yy (xi − µy )(xi − µy )T .
2}. tst = load(’riply_tst’). Example: Fisher Linear Discriminant The binary linear classiﬁer based on the Fisher Linear Discriminant is trained on the Riply’s data set riply trn. it is known that the parameters (µ1 . The parameters (µ1 . Σ2 ) of these class distributions are unknown.tst.y) ans = 0. model = fldqp(trn). However. pline(model). 2} 17 . However.mat. Σi ): i ∈ I2 }.1080 % load training data % compute FLD % % % % plot data and solution load testing data classify testing data compute testing error 2. 2} are known to be multivariate Gaussian distributions. The problem (2.model). The found linear classiﬁer is visualized (see Figure 2. The QP formulation fldqp of FLD is used. ppatterns(trn). The QP solver quadprog of the Matlab Optimization Toolbox is used in the toolbox implementation.4) and evaluated on the testing data riply tst. Σ2 ) belong to a ﬁnite set {(µi .13) can be reformulated as the quadratic programming (QP) task w = argmin w · SW w . ypred = linclass(tst.II. w subject to w · (µ1 − µ2 ) = 2 . Let q: X ⊆ Rn → {1.mat. y ∈ {1. The method is implemented in the function fldqp. Σi ): i ∈ I1 }. This method can be useful when the matrix inversion is hard to compute. Similarly (µ2 .4 Generalized Anderson’s task The classiﬁed object is described by the vector of observations x ∈ X ⊆ Rn and a binary hidden state y ∈ {1. The class conditional distributions pXY (xy). Σ1 ) and (µ2 . Σ1 ) belong to a certain ﬁnite set of parameters {(µi . cerror(ypred. figure. the function fldqp can be replaced by the standard approach implemented in the function fld which would yield the same solution. trn = load(’riply_trn’). References: The Fisher Linear Discriminant is described in Chapter 3 of the book [6] or in the book [3].X.
µi .8 0.b w.2) with discriminant function f (x) = w · x + b. The probability ε(w. 18 . b.4 0..1.2 1 0.6 0. Σi ) is proportional to the reciprocal of the Mahalanobis distance r i between the (µi . b∗ ) is minimal (w ∗ .14) when I1  = 1 and I2  = 1. for f (x) = w∗ · x + b∗ ≥ 0 . i. such that the error Err(w∗ .2 0 −0. µi . b) = max ε(w. q(x) = 2 . In other words. Σi ) .14) The original Anderson’s task is a special case of (2. b) = argmin max ε(w.5 −1 −0.b i∈I1 ∪I2 (2. b∗ ) of the linear classiﬁer 1 . Σi ) is probability that the Gaussian random vector x with mean vector µi and the covariance matrix Σi satisﬁes q(x) = 1 for i ∈ I2 or q(x) = 2 for i ∈ I1 . b∗ ) = argmin Err(w.2 −1. w. µi .4: Linear classiﬁer based on the Fisher Linear Discriminant. Σi ) . Σi ) and the nearest vector of the separating hyperplane H = {x ∈ Rn : w · x + b = 0}.5 1 Figure 2. The probability of misclassiﬁcation is deﬁned as Err(w. b. µi .5 0 0. be a binary linear classiﬁer (2. i∈I1 ∪I2 where ε(w. r i = min (µi − x) · (Σi )−1 (µi − x) = x∈H w · µi + b w · Σi w . for f (x) = w∗ · x + b∗ < 0 . b. The Generalized Anderson’s task (GAT) is to ﬁnd the parameters (w ∗ .e. b. it is the probability that the vector x will be misclassiﬁed by the linear rule q.
b) = argmax min w. 2. The problem is solved by an iterative algorithm which stops while the condition γ− 1−λ ≤ε. the optimal solution vector w ∈ Rn is obtained by solving the following problem w = ((1 − λ)Σ1 + λΣ2 )−1 (µ1 − µ2 ) .5. b) is less than 0. µi . This algorithm solves the original Anderson’s task deﬁned for two Gaussians only I1  = I2  = 1.4. b. A detailed description of the Generalized Anderson’s task and all the methods implemented in the STPRtool is given in book [26].15) The optimization problem (2. In this case. µi . b∗ ) = argmax F (w. b. Σi ) and the corresponding Mahalanobis distance r i is given by the integral ε(w. References: The original Anderson’s task was published in [1]. Example: Solving the original Anderson’s task 19 . 1−λ = λ w · Σ2 w .b w. However. b) is proven to be convex in the region where the probability of misclassiﬁcation Err(w. 2π (2. w · Σi w which is more suitable for optimization. w · Σ1 w where the scalar 0 < λ < 1 is unknown.The exact relation between the probability ε(w. The objective function F (w.14) can be equivalently rewritten as (w ∗ .1 Original Anderson’s algorithm The algorithm is implemented in the function androrig.b i∈I1 ∪I2 w · µi + b . The STPRtool contains implementations of the algorithm solving the original Anderson’s task as well as implementations of three diﬀerent approaches to solve the Generalized Anderson’s task which are described bellow. Σi ) = ∞ ri 1 2 1 √ e− 2 t dt . The parameter ε deﬁnes the closeness to the optimal solution. An interactive demo on the algorithms solving the Generalized Anderson’s task is implemented in demo anderson. w · Σ1 w is satisﬁed. λ γ= w · Σ2 w . the objective function F (w. b) is not diﬀerentiable.
20 .4 gauss 1 0.2 0. The found linear classiﬁer is visualized and the Gaussian distributions (see Figure 2.8 −0.5: Solution of the original Anderson’s task. % figure.X.37 0. trn = load(’riply_trn’). The parameters (µ1 .8 0.7 gauss 2 0.9 1.The original Anderson’s task is solved for the Riply’s data set riply trn.1150 load training data estimate Gaussians solve Anderson’s task plot solution load testing data classify testing data evaluate error 0. Σ2 ) of the Gaussian distribution of the ﬁrst and the second class are obtained by the MaximumLikelihood estimation.6 Figure 2.6 0. distrib ). pandr( model.5). tst.2 0.6 −0.4 −0.5 0.y ) % ans = 0. Σ1 ) and (µ2 . % ypred = linclass( tst. % model = androrig(distrib). % cerror( ypred.4 0.mat.929 0. % distrib = mlcgmm(data). The classiﬁer is validated on the testing set riply tst.2 0 0.3 0.mat.1 −0. model ). % tst = load(’riply_tst’).
Both the ﬁrst and second class are described by three 2Ddimensional Gaussian distributions. b(t+1) ) − F (w (t) . The algorithm iterates until the minimal change in the objective function is less than the prescribed ε. An improving direction (∆w. i. distrib = load(’mars’).. b(t+1) : = b(t) + k∆b .mat. the condition F (w(t+1) .3 εsolution using the Kozinec’s algorithm This algorithm is implemented in the function eanders. The optimal movement k is determined by 1Dsearch based on the cutting interval algorithm according to the Fibonacci series. ∆b).2 General algorithm framework The algorithm is implemented in the function ganders. Example: Solving the Generalized Anderson’s task The Generalized Anderson’s task is solved for the data mars.6).2. The number of interval cuttings of is an optional parameter of the algorithm. model = ganders(distrib). II. b) such that F (w. b(t) ) > ε . pandr(model. is violated.distrib). III. This algorithm repeats the following steps: I. The εsolution is deﬁned as a linear classiﬁer with parameters (w. b) < ε 21 . The found linear classiﬁer as well as the input Gaussian distributions are visualized (see Figure 2.e. ∆b) in the parameter space is computed such that moving along this direction increases the objective function F . The new parameters are set w(t+1) : = w (t) + k∆w . The optimal movement k along the direction (∆w.4.4. The quadprog function of the Matlab optimization toolbox is used to solve this subtask. % load input Gaussians % solve the GAT % plot the solution 2. figure.
The problem is transformed to the task of separation of two sets of ellipsoids which is solved the Kozinec’s algorithm.options).0. The objective function F (w. The found linear classiﬁer and the input Gaussian distributions are visualized (see Figure 2. % % % % load Gaussians maximal desired error training visualization 2. pandr(model.0421 gauss 6 gauss 4 0.0461 gauss 2 gauss 3 2 gauss 1 0. is satisﬁed.4. The desired maximal possible probability of misclassiﬁcation is set to 0. Both the ﬁrst and second class are described by three 2Ddimensional Gaussian distributions.047 0.6: Solution of the Generalized Anderson’s task. eanders(distrib. = struct(’err’.7). = load(’mars’). 0. b) is convex but not diﬀerentiable thus the standard gradient optimization cannot 22 .5 0.5) is the desired upper bound on the probability of misclassiﬁcation.0503 −1 −2 −3 −4 −5 −4 −3 −2 −1 0 1 2 3 4 5 Figure 2. distrib options model = figure.4 Generalized gradient optimization This algorithm is implemented in the function ggradandr.06 (it is 6%).0483 4 3 0.06’). Example: εsolution of the Generalized Anderson’s task The εsolution of the generalized Anderson’s task is solved for the data mars.distrib).mat. Otherwise it can get stuck in an endless loop. The algorithm converges in ﬁnite number of iterations if the given εsolution exists. The prescribed ε ∈ (0.0503 1 0 gauss 5 0.
b(t+1) ) − F (w (t) . i.0473 0. is violated. b(t) ) > ε . the condition F (w(t+1) . Both the ﬁrst and second class are described by three 2Ddimensional Gaussian distributions. The algorithm iterates until the minimal change in the objective function is less than the prescribed ε.0516 −3 0..0496 gauss 1 0. The maximal number of iterations can be used as the stopping condition as well.0483 gauss 5 0 −1 gauss 6 −2 gauss 4 0.e. The optimized parameters are updated w(t+1) : = w (t) + ∆w . II. The maximal number of iterations is limited to tmax=10000. However. The generalized gradient (∆w. ∆b) is computed.7: εsolution of the Generalized Anderson’s task with maximal 0. Example: Solving the Generalized Anderson’s task using the generalized gradient optimization The generalized Anderson’s task is solved for the data mars.8). the generalized gradient optimization method can be applied. b(t+1) = b(t) + ∆b . be used.0516 gauss 3 2 1 0.0432 −4 −5 −4 −3 −2 −1 0 1 2 3 4 5 Figure 2. The algorithm repeats the following two steps: I. 23 .mat.06 probability of misclassiﬁcation. The found linear classiﬁer as well as the input Gaussian distributions are visualized (see Figure 2.5 4 3 gauss 2 0.
% load Gaussians % max # of iterations % training % visualization 5 4 0.0518 gauss 2 gauss 3 3 2 gauss 1 0.0518 −1 0.0475 1 0 gauss 5 0. figure. model = ggradandr(distrib.distrib = load(’mars’).0434 −4 −5 −4 −3 −2 −1 0 1 2 3 4 5 Figure 2. pandr( model. options = struct(’tmax’. 24 . distrib ).8: Solution of the Generalized Anderson’s task after 10000 iterations of the generalized gradient optimization algorithm.10000).0498 0.0485 gauss 6 gauss 4 −2 −3 0.options).
. . The quadratic mapping φ: Rn → Rm is deﬁned as φ(x) = [ x1 .2) . xn . x1 xn . The quadratic data mapping projects the input vector x ∈ Rn onto the feature space Rm where m = n(n + 3)/2. (3.2) is a representative of the supervised learning method yielding the linear projection. . . The projection is given by the matrix W [n × m] and bias vector b ∈ Rm . x1 x1 .3) (3. . x2 .2). . n are entries of the vector x ∈ Rn . The Linear Discriminant Analysis (LDA) (Section 3. i = 1. . .1) is commonly used. 25 (3. . In this case. .1) is a representative of the unsupervised learning method which yields the linear projection (3. . .1). x2 x2 . This yields a nonlinear (kernel) projection of data which has generally deﬁned as z = AT k(x) + b . The kernel functions allow for nonlinear extensions of the linear feature extraction methods. The linear data projection is implemented in function linproj (See examples in Section 3. . . x2 xn .Chapter 3 Feature extraction The feature extraction step consists of mapping the input vector of observations x ∈ Rn onto a new feature description z ∈ Rm which is more suitable for given task. . The quadratic data mapping is implemented in function qmap. . . the input vectors x ∈ Rn are mapped into a new higher dimensional feature space F in which the linear methods are applied. x1 x2 . xn xn ]T . . The linear projection z = WT x + b . where xi .1 and Section 3. The Principal Component Analysis (PCA) (Section 3. The data type used to describe the linear data projection is deﬁned in Table 3. .
1 Table 3. Linear data projection (structure array): . kpcarec Reconstructs image after kernel PCA. The vector k(x) = [k(x. demo pcacomp Demo on image compression using PCA. xl } be a set of training vectors from the ndimensional input space Rn .b [m × 1] Bias vector b. kpca Kernel Principal Component Analysis.1 Principal Component Analysis Let TX = {x1 . . . .3) and the Generalized Discriminant Analysis (GDA) (Section 3.where A [l × m] is a parameter matrix and b ∈ Rm a parameter vector. . The kernel data projection is implemented in the function kernelproj (See examples in Section 3. . qmap Quadratic data mapping. The extracted features are projections of φ(x) onto m vectors (or functions) from the feature space F . pca Principal Component Analysis. The x ∈ Rn stands for the input vector and z ∈ Rm is the vector of extracted features. 3. lin2quad Merges linear rule and quadratic mapping. . The data type used to describe the kernel data projection is deﬁned in Table 3. k(x. wm ]. .5). . pcarec Computes reconstructed vector after PCA projection. . The summary of implemented methods for feature extraction is given in Table 3. z l } is a lower dimensional representation of the 26 .5) are representatives of the feature extraction methods yielding the kernel data projection (3. The set of vectors TZ = {z 1 . lin2svm Merges linear rule and kernel projection. kerproj Kernel data projection. gda Generalized Discriminant Analysis.3 and Section 3. x1 ). . xl )]T is a vector of kernel functions centered in the training data or their subset. . . .fun = ’linproj’ Identiﬁes function associated with this data type. . . . lda Linear Discriminant Analysis. . Table 3.1: Implemented methods for feature extraction linproj Linear data projection.W [n × m] Projection matrix W = [w1 . greedykpca Greedy Kernel Principal Component Analysis. .3).2: Datatype used to describe linear data projection. The Kernel PCA (Section 3.
i = 1. m are column vectors of the matrix W = [w1 . . b) = l l ˜ xi − xi i=1 2 . xl }. input training vectors TX in the mdimensional space Rm .b (3. . . where µ is the sample mean of the training data. b) = argmin εM S (W . . .4). . wm ] and δ(i. j) is the Kronecker delta function. j) .Alpha [l × m] Parameter matrix A [l × m].3: Datatype used to describe kernel data projection.7) subject to wi · wj = δ(i. W .5). The vectors TZ are obtained by the linear orthonormal projection z = WT x + b . 27 .options.7) is the matrix W = [w 1 .4) where the matrix W [n × m] and the vector b [m × 1] are parameters of the projec˜ ˜ tion.sv. The Principal Component Analysis (PCA) is the linear orthonormal projection (3. w m ] containing the m eigenvectors of the sample covariance matrix which have the largest eigen values.6) is a function of the parameters of the linear projections (3. (3.Table 3.6) of the training data TX .4) which allows for the minimal mean square reconstruction error (3. ∀i. xl } are computed by the linear back ˜ projection ˜ x = W(z − b) . . The PCA is implemented in the function pca. . . The mean square reconstruction error 1 εM S (W. .X [n × l] Training vectors {x1 . The parameters (W. The solution of the task (3.4) and (3. (3. b ) . . (3. . where w i .5) obtained by inverting (3. . . . .b [m × 1] Bias vector b.options.fun = ’kernelproj’ Identiﬁes function associated with this data type. .arg [1 × p] Kernel argument(s). . A demo showing the use of PCA for image compression is implemented in the script demo pcacomp. . . Kernel data projection (structure array): . b) of the linear projection are the solution of the optimization task (W. . . The vector b equals to WT µ.ker [string] Kernel identiﬁer. . j . . . The reconstructed vectors TX = {x1 .
% visualization h1 = ppatterns(X. Z = linproj(X.1: Example on the Principal Component Analysis.5 6 5. axis equal.8 1]. 3. ’Input vectors’. XR = pcarec(X. model = pca(X.’kx’).mx] = max(Z).5 5 4.mx)].model). data figure. % % % % generate data train PCA lower dim. [dummy.0.5 4 3.5 3 2.mn] = min(Z).’r’).References: The PCA is treated throughout for instance in books [13. The training data. proj.mx)]. [dummy. h2 = ppatterns(XR.. 6]. h3 = plot([XR(1.[XR(2.mn) XR(1.8. 28 . the reconstructed data and the 1dimensional subspace given by the ﬁrst principal component are visualized (see Figure 3.1). 7. X = gsamp([5.[1 0. ’PCA subspace’). legend([h1 h2 h3].1).mn) XR(2. reconstr. Example: Principal Component Analysis The input data are synthetically generated from the 2dimensional Gaussian distribution.’bo’).model). The PCA is applied to train the 1dimensional subspace to approximate the input data with the minimal reconstruction error.5 7 6.’Reconstructed’.5].100).5 2 3 4 5 6 7 Input vectors Reconstructed PCA subspace 8 Figure 3. . hold on..
. figure. 6].8). model = lda(orig_data. . Iy  i∈I y The goal of the Linear Discriminant Analysis (LDA) is to train the linear data projection z = WT x . ext_data = linproj(orig_data.3. (xl . y ∈ Y = {1. % % % % load input data train LDA feature extraction plot ext. i=1 µy = 1 xi . . .model). .mat which consists of the labeled 4dimensional data.8) SB = y∈Y Iy (µy − µ)(µy − µ)T . such that the class separability criterion F (W) = ˜ det(SB ) det(SB ) . 2. The data after the feature extraction step are visualized in Figure 3. . c} be labeled set of training vectors. . y1). The SB is the betweenclass and SW is the withinclass scatter matrix of projected data which is deﬁned similarly to (3. (3. y ∈ Y . where the total mean vector µ and the class mean vectors µy .2). Sy = i∈Iy (xi − µi )(xi − µi )T . orig_data = load(’iris’). Example 1: Linear Discriminant Analysis The LDA is applied to extract 2 features from the Iris data set iris.2. data Example 2: PCA versus LDA This example shows why the LDA is superior to the PCA when features good for classiﬁcation are to be extracted. = ˜ det(SW ) det(SW ) ˜ ˜ is maximized.2 Linear Discriminant Analysis Let TXY = {(x1 . yl )}. ppatterns(ext_data). xi ∈ Rn . The synthetical data are generated from the Gaussian mixture model. The withinclass SW and betweenclass SB scatter matrices are described as SW = y∈Y Sy . The LDA and PCA are trained on the generated data. . References: The LDA is treated for instance in books [3. The LDA and 29 . y ∈ Y are deﬁned as 1 µ= l l xi .
5]].2: Example shows 2dimensional data extracted from the originally 4dimensional Iris data set using Linear Discriminant Analysis.5 0.lda_model). ppatterns(data). pca_data = linproj( data.lda_model). lda_data = linproj(data.1).1) = [1 0.5 1 1. lda_rec = pcarec(data.3 0.5].:.Cov(:. distrib.9.Mean = [[5.2 −0. 0. lda_model = lda(data.0. 0.pca_model).9.4 −0. 30 % % % % % mean vectors 1st covariance 2nd covariance Gaussian weights sample data % train LDA % train PCA % visualization .1 −0.2) = [1 0. figure.3a). PCA directions are displayed as connection of the vectors projected to the trained LDA and PCA subspaces and reprojected back to the input space (see Figure 3.2 0. The extracted data using the LDA and PCA model are displayed with the Gaussians ﬁtted by the MaximumLikelihood (see Figure 3.pca_model). distrib.5 −1 −0.1).5 0 0.3b).5 −1.4] [4.X.Cov(:.:.3 −0.5 0.250). distrib. pca_rec = pcarec(data. data = gmmsamp(distrib. hold on.9 1]. pca_model = pca(data.4 0. % Generate data distrib.X.9 1].1 0 −0. axis equal.Prior = [0.X.5 Figure 3.
. pgauss(mlcgmm(pca_data)). β = A(z − b) . hold on. .1.10) ˜ is minimized.1. . title(’LDA’).’b’). (3.pca_rec(2. The problem is to ﬁnd the vector x ∈ X its image ˜ φ(x) ∈ F well approximates the reconstructed vector φ(x) ∈ F . b) = l l (3. such that the reconstruction error 1 εKM S (A.9) ˜ φ(xi ) − φ(xi ) i=1 2 . . ppatterns(pca_data). The input training vectors TX = {x1 . Project input vector xin ∈ Rn onto its lower dimensional representation z ∈ Rm using (3.’r’). title(’PCA’). The linear PCA is applied on the mapped data TΦ = {φ(x1 ).9). subplot(2.h1 = plot(lda_rec(1. The reconstructed vector φ(x) is given as a linear combination of the mapped data TΦ l ˜ φ(x) = i=1 βi φ(xi ) . This procedure is implemented in the function kpcarec for the case of the Radial Basis Function (RBF) kernel.:). . pgauss(mlcgmm(lda_data)).:). .2). The kernel PCA trains the kernel data projection z = AT k(x) + b .lda_rec(2. subplot(2.11) In contrast to the linear PCA.’PCA direction’). legend([h1 h2]. 31 . (3.1). 3. figure.3 Kernel Principal Component Analysis The Kernel Principal Component Analysis (Kernel PCA) is the nonlinear extension of the ordinary linear PCA.:).:). h2 = plot(pca_rec(1. ppatterns(lda_data). . xi ∈ X ⊆ Rn are mapped by φ: X → F to a high dimensional feature space F .’LDA direction’. The procedure consists of the following step: I. the explicit projection from the feature space F to the input space X usually do not exist. . xl }. The computation of the principal components and the projection on these components can be expressed in terms of dot products thus the kernel functions k: X × X → R can be employed. φ(xl )}.
The function kpcarec is used to ﬁnd preimages of the reconstructed data (3.5. . xi ∈ Rn be the set of input training vectors.11). . The result is visualized in Figure 3. h2 = ppatterns(XR. (3. The input data are synthetically generated 2dimensional vectors which lie on a circle and are corrupted by the Gaussian noise. b [m × 1] is the parameter vector and kS = [k(x.’Reconstructed’). In contrast to the ordinary kernel PCA.model). % Visualization h1 = ppatterns(X). % compute reconstruced data figure.9). Let TX = {x1 . % use RBF kernel options. xl }.’Input vectors’. ’+r’). % generate circle data options. % output dimension model = kpca(X. Computes vector xout ∈ Rn which is good preimage of the reconstructed vector ˜ φ(xin ). . k(x.1). The objective of the Greedy kernel PCA is to minimize the reconstruction error (3. 3.arg = 4. The Greedy kernel PCA algorithm is implemented in the function greedykpca.II. legend([h1 h2]. .ker = ’rbf’.options). 32 .11) while the size d of the subset TS is kept small.1]. . . .4 Greedy kernel PCA algorithm The Greedy Kernel Principal Analysis (Greedy kernel PCA) is an eﬃcient algorithm to compute the ordinary kernel PCA. X = gencircledata([1. Example: Kernel Principal Component Analysis The example shows using the kernel PCA for data denoising. The goal is to train the kernel data projection z = AT kS (x) + b . 30].12) is reduced compared to (3.4. . % compute kernel PCA XR = kpcarec(X. such that ˜ xout = argmin φ(x) − φ(xin ) 2 . s1 ). % kernel argument options. x References: The kernel PCA is described at length in book [29. .12) where A [d × m] is the parameter matrix.250.new_dim = 2. sd }. . . The vector set TS is a subset of training data TX . the subset TS does not contain all the training vectors TX thus the complexity of the projection (3. The kernel PCA model with RBF kernel is trained. . sl )]T are the kernel functions centered in the vectors TS = {s1 .
..3) requires all the 250 training vectors. (xl . yl )}. The vector of the selected subset TX are visualized as well. The resulting kernel data projection is deﬁned as z = AT k(x) + b . . . The function kpcarec is used to ﬁnd preimages of the reconstructed data obtained by the kernel PCA projection. 3.. h2 = ppatterns(XR. % generate training data options.12).12) is limited to 25.1]. model = greedykpca(X. y ∈ Y = {1. The input data are synthetically generated 2dimensional vectors which lie on a circle and are corrupted with the Gaussian noise.options). (φ(xl ). The Greedy kernel PCA model with RBF kernel is trained.’+r’). y1 ). options.new_dim = 2.5 Generalized Discriminant Analysis The Generalized Discriminant Analysis (GDA) is the nonlinear extension of the ordinary Linear Discriminant Analysis (LDA). . 33 . .arg = 4. options. The ordinary LDA is applied on the mapped data TΦY = {(φ(x1 ).5.model). h3 = ppatterns(model.sv. % reconstructed vectors figure.’ob’. options.X. xi ∈ X ⊆ Rn . % visualization h1 = ppatterns(X).1). yl )}. % % % % % use RBF kernel kernel argument output dimension size of sel. . . .ker = ’rbf’.’Selected subset’).m = 25. Example: Greedy Kernel Principal Component Analysis The example shows using kernel PCA for data denoising. .250.References: The implemented Greedy kernel PCA algorithm is described in [11]. X = gencircledata([1. legend([h1 h2 h3]. y1). . The input training data TXY = {(x1 . ’Training set’. . The training and reconstructed vectors are visualized in Figure 3. subset run greedy algorithm XR = kpcarec(X.’Reconstructed set’. c} are mapped by φ: X → F to a high dimensional feature space F . . . The number of vectors deﬁning the kernel projection (3. In contrast the ordinary kernel PCA (see experiment in Section 3. The computation of the projection vectors and the projection on these vectors can be expressed in terms of dot products thus the kernel functions k: X × X → R can be employed.5.
that setting σ too low leads to the perfect separability of the training data TXY but the kernel projection is heavily overﬁtted. model = gda(orig_data.6 Combining feature extraction and linear classiﬁer Let TXY = {(x1 . c}. The GDA is implemented in the function gda. 34 (3. output dimension train GDA feature ext. In the case of the quadratic data mapping (3. . z i ∈ Rm . . options. . yl )}. quadratic data mapping or kernel projection). The RBF kernel with kernel width σ = 1 is used. . . . visualization 3. Notice. . (φ(xl ).arg = 1. . the discriminant function can be written in the following form f (x) = x · Ax + x · b + c . ext_data = kernelproj(orig_data. . . figure. (xl .options). .mat which contains 3 classes of 4dimensional data. y1 ). .6. . b) are trained to increase the betweenclass scatter and decrease the withinclass scatter of the extracted data TZY = {(z 1 . The parameters (A. Let φ: X → Rm be a mapping φ(x) = [φ1 (x). options. orig_data = load(’iris’). options. φm (x)]T deﬁning a nonlinear feature extraction step (e.model). Let f (x) = w · φ(x) + b be a linear discriminant function found by an arbitrary training method on the extracted data TΦY . yl )} be a training set of observations xi ∈ X ⊆ Rn and corresponding hidden states yi ∈ Y = {1. ppatterns(ext_data). . .new_dim = 2. The feature extraction step applied to the input data TXY yields extracted data TΦY = {(φ(x1 ). . The function f can be seen as a nonlinear function of the input vectors x ∈ X . % % % % % % % load data use RBF kernel kernel arg. (z l .where A [l × m] is the matrix and b [m × 1] the vector trained by the GDA. . y1 ). . Example: Generalized Discriminant Analysis The GDA is trained on the Iris data set iris.13) . The extracted data YZY are visualized in Figure 3.ker = ’rbf’.2) used as the feature extraction step. .. yl )}. References: The GDA was published in the paper [2]. . as m f (x) = w · φ(x) + b = i=1 wiφi (x) + b .g. y1 ).
The Perceptron algorithm produces the linear rule only. ppatterns(input_data). pboundary(quad_model). Function lin2quad is used to compute the parameters of the quadratic classiﬁer explicitly based on the found linear rule in the feature space and the mapping.7 shows the decision boundary of the quadratic classiﬁer.3) applied as the feature extraction step. In the case of the kernel projection (3. The input training data are linearly nonseparable but become separable after the quadratic data mapping is applied.14) deﬁnes the kernel (SVM type) classiﬁer described in Section 5. % map data to the feature space map_data = qmap(input_data).13) deﬁnes the quadratic classiﬁer described in Section 6. (ii) training a linear rule on the extracted data and (iii) combing the kernel projection and the linear rule. The 35 . Therefore the quadratic classiﬁer can be trained by (i) applying the quadratic data mapping qmap. % train linear rule in the feature sapce lin_model = perceptron(map_data). The combining is implement in function lin2quad. (3.14) The discriminant function (3.The discriminant function (3.4. This classiﬁer can be trained by (i) applying the kernel projection kernelproj. % visualize the quadratic classifier figure. Example: Quadratic classiﬁer trained the Perceptron This example shows how to train the quadratic classiﬁer using the Perceptron algorithm. perceptron. (ii) training a linear rule on the mapped data (e. Example: Kernel classiﬁer from the linear rule This example shows how to train the kernel classiﬁer using arbitrary training algorithm for linear rule. The Fisher Linear Discriminant was used in this example. the resulting discriminant function has the following form f (x) = α · k(x) + b . % load training data living in the input space input_data = load(’vltava’). % compute parameters of the quadratic classifier quad_model = lin2quad(lin_model).. The combining is implemented in function lin2svm. fld) and (iii) combining the quadratic mapping and the linear rule. Figure 3.g.
8 shows the found kernel classiﬁer. ppatterns(kfd_model. cerror(ypred.’arg’.y) ans = 0.1. ypred = svmclass(tst.lin_model).X. % project data map_trn = kernelproj(trn. the training data and vectors used in the kernel expansion of the discriminant function.kfd_model). Figure 3.sv. pboundary(kfd_model). The Riply’s data set was used for training and evaluation.X.10).’rbf’.tst. % combine linear rule and kernel PCA kfd_model = lin2svm(kpca_model.13). % visualization figure. % load input data trn = load(’riply_trn’).’new_dim’. % train linear classifier lin_model = fld(map_trn).kpca_model). % evaluation on testing data tst = load(’riply_tst’).options). kpca_model = greedykpca(trn.0970 36 . % train kernel PCA by greedy algorithm options = struct(’ker’.greedy kernel PCA algorithm was applied as the feature extraction step.’ok’. ppatterns(trn).X.
7
LDA direction PCA direction
6
5
4
3
2
1
0
1
2
3
4
5
6
7
8
(a)
LDA 1.4 1.2 1 0.8 0.6 0.4 0.2 0 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2
PCA 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 −4 −3 −2 −1 0 1 2 3 4 5
(b) Figure 3.3: Comparison between PCA and LDA. Figure (a) shows input data and found LDA (red) and PCA (blue) directions onto which the data are projected. Figure (b) shows the class conditional Gaussians estimated from data projected data onto LDA and PCA directions.
37
10 Input vectors Reconstructed 8
6
4
2
0
−2
−4
−6
−8 −8
−6
−4
−2
0
2
4
6
8
10
Figure 3.4: Example on the Kernel Principal Component Analysis.
8 Training set Reconstructed set Selected set 6
4
2
0
−2
−4
−6 −8
−6
−4
−2
0
2
4
6
8
10
Figure 3.5: Example on the Greedy Kernel Principal Component Analysis.
38
0.15
0.1
0.05
0
−0.05
−0.1
−0.15
−0.2 −0.6
−0.5
−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
Figure 3.6: Example shows 2dimensional data extracted from the originally 4dimensional Iris data set using the Generalized Discriminant Analysis. In contrast to LDA (see Figure 3.2 for comparison), the class clusters are more compact but at the expense of a higher risk of overﬁtting.
0.8
0.6
0.4
0.2
0
−0.2
−0.4
−0.6 −0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Figure 3.7: Example shows quadratic classiﬁer found by the Perceptron algorithm on the data mapped to the feature by the quadratic mapping.
39
1.4 0.2 −1. The greedy kernel PCA was used therefore only a subset of training data was used to determine the kernel projection.5 0 0.6 0.8: Example shows the kernel classiﬁer found by the Fisher Linear Discriminant algorithm on the data mapped to the feature by the kernel PCA.5 1 Figure 3. 40 .8 0.2 0 −0.2 1 0.5 −1 −0.
1 Gaussian distribution and Gaussian mixture model The value of the multivariate Gaussian probability distribution function in the vector x ∈ Rn is given by PX (x) = 1 (2π) n 2 1 exp(− (x − µ)T Σ−1 (x − µ)) .2).Chapter 4 Density estimation and clustering The STPRtool represents a probability distribution function PX (x). Σi) is assigned to the class yi ∈ Y. . The STPRtool uses speciﬁc datatype to describe a set of Gaussians which can be labeled. . xl }. The vector y [1 × l] contains values of the distribution function.4. the shape of the matrix can be indicated in the optimal ﬁeld . however. . In particular.e. The datatype is described in Table 4.. 2 det(Σ) (4. . Let the structure model represent a given distribution function.1. The Matlab command y = feval(X. x ∈ Rn as a structure array which must contain ﬁeld . the Gaussian distribution and the Gaussian mixture models are described in Section 4.1) where µ [n × 1] is the mean vector and Σ [n × n] is the covariance matrix. . The summary of implemented methods for dealing with probability density estimation and clustering is given in Table 4. i.3). xl } of a random vector. The matrix X [n × l] contains column vectors {x1 . .fun) is used to evaluate yi = PX (xi ).1 4. . each pair (µi . The ﬁeld .fun speciﬁes function which is used to evaluate PX (x) for given set of realization {x1 .cov type (see Table 4. . The covariance matrix Σ is always represented as the square matrix [n × n]. The evaluation of the Gaussian distribution is implemented in the function 41 .model.fun (see Table 4.
. . Probability Distribution Function (structure array): .fun [string] The name of a function which evaluates probability distribution y = feval( X.2) where PY (y).2: Datatype used to represent probability distribution function. y ∈ Y = {1.fun ). Table 4. The datatype used to represent the GMM is described in Table 4. gsamp Generates sample from Gaussian distribution. demo svmpout Fitting a posteriori probability to SVM output. melgmm Maximizes Expectation of LogLikelihood for Gaussian mixture. The evaluation of the GMM is implemented in the function pdfgmm. . pdfgauss. The Gaussian Mixture Model (GMM) is weighted sum of the Gaussian distributions PX (x) = y∈Y PY (y)PXY (xy) .Table 4. sigmoid Evaluates sigmoid function. kmeans Kmeans clustering algorithm. (4. mlsigmoid Fitting a sigmoid function using ML estimation. The sampling from univariate and bivariate Gaussians is shown. demo emgmm Demo on ExpectationMaximization (EM) algorithm. model. rsde Reduced Set Density Estimator. Example: Sampling from the Gaussian distribution The example shows how to represent the Gaussian distribution and how to sample data it. mlcgmm Maximal Likelihood estimation of Gaussian mixture model. c} are weights and PXY (xy) is the Gaussian distribution (4. pdfgauss Evaluates Gaussian mixture model.1: Implemented methods: Density estimation and clustering gmmsamp Generates sample from Gaussian mixture model. Σy ). pdfgauss Evaluates for multivariate Gaussian distribution. mmgauss Minimax estimation of Gaussian distribution. .5. demo mmgauss Demo on minimax estimation for Gaussian. emgmm ExpectationMaximization Algorithm for Gaussian mixture model. The distribution function of the univariate Gaussian is visualized and compared to the histogram 42 .1) given by parameters (µy . The random sampling from the GMM is implemented in the function gmmsamp. The random sampling from the Gaussian distribution is implemented in the function gsamp.
Cov = [1 0. Σk }. % Mean vector % Covariance matrix % plot sampled data % plot shape 4. . of the sample data (see Figure 4. .2 MaximumLikelihood estimation of GMM PXY Θ (x.’r’).X] = hist(gsamp(model. Labeled set of Gaussians (structure array): . .Table 4. 0.3).3: Shape of covariance matrix.6. . hold on. . . % Gaussian parameters figure.Mean [n × k] Mean vectors {µ1 .cov type [string] Optional ﬁeld which indicates shape of the covariance matrix (see Table 4. . . Parameter cov type: ’full’ Full covariance matrix.1:5]. . model. .1a).10). The isoline of the distribution function of the bivariate Gaussian is visualized together with the sample data (see Figure 4. θ)PY Θ (yθ) . of the Gaussisan plot([4:0. µk }.2). . yk } assigned to Gaussians. hold on. ’spherical’ Spherical covariance matrix. [Y.y [1 × k] Labels {y1 . . % univariate case model = struct(’Mean’. . % plot pdf.Y/500). pgauss(model).1b). . Table 4. figure. ’diag’ Diagonal covariance matrix. yθ) = PXY Θ (xy.1:5]. .fun = ’pdfgauss’ Identiﬁes function associated with this data type.6 1]. 43 Let the probability distribution .Cov [n × n × k] Covariance matrices {Σ1 .1]. .500).1.pdfgauss([4:0. % plot histogram % bivariate case model.’Cov’.4: Datatype used to represent the labeled set of Gaussians.250)). ppatterns(gsamp(model. % compute histogram bar(X.model).Mean = [1.
. µc }.d.2). The pairs (xi . . PK (c)}. yi) are assumed to samples from random variables which are independent and identically distributed (i. 4.2.2 1 0.2. Thanks to the 44 . The marginal probability PXΘ (xθ) is the Gaussian mixture model (4. yθ). .fun = ’pdfgmm’ Identiﬁes function associated with this data type.1 −1 0. . Σy ). . . The θ involves parameters (µy . y ∈ Y. . . .1: Sampling from the univariate (a) and bivariate (b) Gaussian distribution.2.35 5 4 0. yl )} which contains both vectors of observations xi ∈ Rn and corresponding hidden states yi ∈ Y. .1 Complete data The input is a set TXY = {(x1 . . y ∈ Y of Gaussian components and the values of the discrete distribution PY Θ (yθ). . θ) is assumed to be Gaussian distribution. .Table 4. The MaximumLikelihood estimation of the parameters θ of the GMM for the case of complete and incomplete data is described in Section 4. Gaussian Mixture Model (structure array): . . Σc }. . (xl .5: Datatype used to represent the Gaussian Mixture Model. . .15 gauss 1 0. . be known up to parameters θ ∈ Θ. . . . 0.) according to PXY Θ (x.05 −2 0 −4 −3 −2 −1 0 1 2 3 4 5 6 −3 −2 −1 0 1 2 3 4 5 (a) (b) Figure 4. c} are assumed to be realizations of random variables which are independent and identically distributed according to the PXY Θ (x.25 2 0. The logarithm of probability P (TXY θ) is referred to as the loglikelihood L(θTXY ) of θ with respect to TXY . . The vector of observations x ∈ Rn and the hidden state y ∈ Y = {1.127 0 0.i.Mean [n × c] Mean vectors {µ1 .3 3 0. respectively.1 and Section 4. .Cov [n × n × c] Covariance matrices {Σ1 .Prior [c × 1] Weights of Gaussians {PK (1). yθ).2. y1). The conditional probability distribution PXY Θ (xy.
45 .i. hold on. ppatterns(data).3) The ML estimate of the parameters θ of the GMM from the complete data set TXY has an analytical solution. hold on. data = load(’riply_trn’).2 Incomplete data The input is a set TX = {x1 .i. Thanks to the i.i. assumption the loglikelihood can be factorized as l L(θTXY ) = log P (TXY θ) = i=1 log PXY Θ (xi . assumption the loglikelihood can be factorized as l L(θTX ) = log P (TX θ) = i=1 y∈Y PXY Θ (xi yi. yiθ) . figure. . xl } which contains only vectors of observations xi ∈ Rn without knowledge about the corresponding hidden states yi ∈ Y.’contour’)).2b) . pgauss(model).d. .d. 4.struct(’visual’. The computation of the ML estimate is implemented in the function mlcgmm. Example: MaximumLikelihood estimate from complete data The parameters of the Gaussian mixture model is estimated from the complete Riply’s data set riply trn. ppatterns(data).2. . % ML estimate of GMM % visualization figure. The logarithm of probability P (TX θ) is referred to as the loglikelihood L(θTX ) of θ with respect to TX . % load labeled (complete) data model = mlcgmm(data).2a) and (ii) contours of the distribution function (see Figure 4. yiθ) . The estimated GMM is visualized as: (i) shape of components of the GMM (see Figure 4. The data are binary labeled thus the GMM has two Gaussians components. θ)PY Θ (yi θ) . The problem of MaximumLikelihood (ML) estimation from the complete data TXY is deﬁned as l θ = argmax L(θTXY ) = argmax θ∈Θ θ∈Θ i=1 ∗ log PXY Θ (xi . (4. pgmm(model. .
46 . true_model.3a. The random initialization is used by default. i. such that the loglikelihood L(θ (t) TX ) monotonically increases. . .The problem of MaximumLikelihood (ML) estimation from the incomplete data TX is deﬁned as l θ = argmax L(θTX ) = argmax θ∈Θ θ∈Θ i=1 y∈Y ∗ PXY Θ (xi yi.5]. A nice description can be found in book [3]. The numerical optimization has to be used instead.4) The ML estimate of the parameters θ of the GMM from the incomplete data set TXY does not have an analytical solution. The ground truth model. 250). Example: MaximumLikelihood estimate from incomplete data The estimation of the parameters of the GMM is demonstrated on a synthetic data.Mean = [2 2]. L(θ (0) TX ) < L(θ (1) TX ) < . The ﬁrst papers describing the EM are [25. θ (t) . < L(θ (t) TX ) until a stationary point L(θ (t−1) TX ) = L(θ (t) TX ) is achieved. The EM algorithm builds a sequence of parameter estimates θ (0) . The ExpectationMaximization (EM) algorithm is an iterative procedure for ML estimation from the incomplete data.4 0. . The monograph [18] is entirely devoted to the EM algorithm.Cov = [1 0. sample = gmmsamp(true_model. The random guess or Kmeans algorithm (see Section 4. . The EM algorithm is applied to estimate the GMM parameters.Prior = [0. References: The EM algorithm (including convergence proofs) is described in book [26]. θ)PY Θ (yi θ) . . An interactive demo on the EM algorithm for the GMM is implemented in demo emgmm. The EM algorithm for the GMM is implemented in the function emgmm. The ground truth model is sampled to obtain a training (incomplete) data. (4. true_model. The ground truth model is the GMM with two univariate Gaussian components. % ground truth Gaussian mixture model true_model. . 5]. The plot of the loglikelihood function L(θ (t) TX ) with respect to the number of iterations t is visualized in Figure 4..6]. The EM algorithm is a local optimization technique thus the estimated parameters depend on the initial guess θ (0) . sampled data and the estimate model are visualized in Figure 4. θ (1) .5) are commonly used to select the initial θ (0) .3b.e.
. .X). x∈TX (4.1). The vectors TX are assumed to be a realizations with high value of the distribution function PXΘ (xθ). References: The minimax algorithm for probability estimation is described and analyzed in Chapter 8 of the book [26]. 4. . h1 = pgmm(true_model. xlabel(’iterations’). The algorithm is implemented in the function mmgauss. 47 . .X. h2 = pgmm(estimated_model.% ML estimation by EM options. . The inequality (4. The minimax estimation is deﬁned as θ ∗ = argmax min log PXΘ (xθ) . θ∈Θ x∈TX (4. % visualization figure. The algorithm converges in a ﬁnite number of iterations to such an estimate θ that the condition x∈TX min log PXΘ (xθ ∗ ) − min log PXΘ (xθ) < ε . .struct(’color’. legend([h1(1) h2(1)]. An interactive demo on the minimax estimation for the Gaussian distribution is implemented in the function demo mmgauss.ncomp = 2. % display progress info estimated_model = emgmm(sample.6) holds for ε > 0 which deﬁnes closeness to the optimal solution θ ∗ .options).’b’)).struct(’color’.3 Minimax estimation The input is a set TX = {x1 .6) can be evaluated eﬃciently thus it is a natural choice for the stopping condition of the algorithm. The PXΘ (xθ) is known up to the parameter θ ∈ Θ. ylabel(’loglikelihood’). The algorithm builds a series of estimates θ (0) .’r’)). θ (1) . hold on. ppatterns(sample. xl } contains vectors xi ∈ Rn which are realizations of random variable distributed according to PXΘ (xθ).5) If a procedure for the MaximumLikelihood estimate (global solution) of the distribution PXΘ (xθ) exists then a numerical iterative algorithm which converges to the optimal estimate θ ∗ can be constructed. θ (t) which converges to the optimal estimate θ ∗ . ’ML estimation’). .verb = 1.logL ). figure. plot( estimated_model. . The STPRtool contains an implementation of the minimax algorithm for the Gaussian distribution (4.’Ground truth’. % number of Gaussian components options.
0] [1.4 Probabilistic output of classiﬁer In general. struct(’p’. The found solution is visualized as the isoline of the distribution function at the point PXΘ (xθ) (see Figure 4. θ) = 1 . . The parameters θ = [a1 .g.4). the probabilistic output of the classiﬁer can help in postprocessing. Let TXY = {(x1 .Example: Minimax estimation of the Gaussian distribution The bivariate Gaussian distribution is described by three samples which are known to have high value of the distribution function. .0] [0.f. . θ) of the hidden state y given the value of the discriminant function f (x). In this particular case. (xl . Perceptron). The aim is to estimate parameters of a posteriori distribution PY F Θ (yf (x). in combining more classiﬁers together. ppatterns(X. . 4. the discriminant function of an arbitrary classiﬁer does not have meaning of probability (e. yl )} is the training set composed of the vectors xi ∈ X ⊆ Rn and the corresponding binary hidden states yi ∈ Y = {1.exp(model. pgauss(model. % points with high p. SVM. a2 ]T . y1 ). the minimax problem is equivalent to searching for the minimal enclosing ellipsoid.d. % run minimax algorithm % visualization figure. θ) . The loglikelihood function is deﬁned as l L(θTXY ) = i=1 log PY F Θ (yif (xi ). 2}. X = [[0. θ ) . The distribution is modeled by a sigmoid function PY F Θ (1f (x).1]]. (4. The training set TXY is assumed to be identically and independently distributed from underlying distribution..13). 1 + exp(a1 f (x) + a2 ) which is determined by parameters θ = [a1 . a2 ]T can be estimated by the maximumlikelihood method l θ = argmax L(θ TXY ) = argmax θ θ i=1 log PY F Θ (yi f (xi ). Fitting a sigmoid function to the classiﬁer output is a way how to solve this problem.7) 48 .’xr’. model = mmgauss(X).lower_bound))). The minimax algorithm is applied to estimate the parameters of the Gaussian. for instance. However. Let f : X ⊆ Rn → R be a discriminant function trained from the data TXY by an arbitrary learning method.
The example is implemented in the script demo svmpout. The GMM allows to compute the a posteriori probability for comparison to the sigmoid estimate. % fit sigmoid function to svm output sigmoid_model = mlsigmoid(svm_output). a2 ]T of the sigmoid function.svm_output. Example: Probabilistic output for SVM This example is implemented in the script demo svmpout. hold on.1.6: Datatype used to describe sigmoid function.y = data.X.X] = svmclass(data.y. . svm_model = smo(data. % train SVM options = struct(’ker’. Figure 4. The evaluation of the sigmoid is implemented in function sigmoid.options).’C’. % visulization figure.5b shows the decision boundary of SVM classiﬁer for illustration. The function produces the parameters θ = [a1 . The ML estimated of Gaussian mixture model (GMM) of the SVM output is used for comparison. Table 4. References: The method of ﬁtting the sigmoid function is described in [22]. 49 .7). % ML estimation of GMM model of SVM output gmm_model = mlcgmm(svm_output).’arg’. Sigmoid function (structure array): .10). The Riply’s data set riply trn is used for training the SVM and the ML estimation of the sigmoid. % compute SVM output [dummy. Figure 4. a2 ]T which are stored in the datatype describing the sigmoid (see Table 4.5a shows found probabilistic models of a posteriori probability of the SVM output. svm_output.’rbf’.A [2 × 1] Parameters θ = [a1 .fun = ’sigmoid’ Identiﬁes function associated with this data type.svm_model).6).The STPRtool provides an implementation mlsigmoid which uses the Matlab Optimization toolbox function fminunc to solve the task (4. % load training data data = load(’riply_trn’). The example shows how to train the probabilistic output for the SVM classiﬁer.
ppatterns(svm_output).’P(y=1f(x)) MLSigmoid’.c l i=1 θ θ The Kmeans1 is an iterative procedure which iteratively builds a series θ (0) .X).’g’). . . θ (1) . . θ = argmin F (θ) = argmin y=1.c The small value of F (θ) indicates that the centers θ well describe the input set TX . sigmoid_apost = sigmoid(fx. xl } which contains vectors xi ∈ Rn .sigmoid_apost.e.Prior(2))). hgmm = plot(fx. .hcomp]..:)*gmm_model.’P(f(x)y=1) MLGMM’.sigmoid_model). % GMM model pcond = pdfgauss( fx. .. Let the objective function F (θ) be deﬁned as l 1 F (θ) = min µy − xi 2 .Prior(1)).. legend([hsigmoid.xlabel(’svm output f(x)’). . ylabel(’p(y=1f(x))’).:)*gmm_model. .. % SVM decision boundary figure. gmm_model)...5 Kmeans clustering The input is a set TX = {x1 . θ (t) . 4. the task is to solve l 1 ∗ min µy − xi 2 . (pcond(1. µc }. l i=1 y=1.’P(f(x)y=2) MLGMM’). hcomp = pgauss(gmm_model). hsigmoid = plot(fx. i./.X). µi ∈ Rn be a set of unknown cluster centers. max(svm_output. . % sigmoid model fx = linspace(min(svm_output. ’P(y=1f(x)) MLGMM’. gmm_apost = (pcond(1..hgmm. 200). The Kmeans algorithm converges in a ﬁnite number of steps to the local optimum of the objective function F (θ).:)*gmm_model. Let θ = {µ1 .’k’). ..gmm_apost.Prior(1)+(pcond(2.. . The found solution depends on the initial guess θ (0) which 1 The K in the name Kmeans stands for the number of centers which is here denoted by c. ... ppatterns(data). . psvm(svm_model).. The task is to ﬁnd such centers θ which describe the set TX best.. 50 ..
% load data [model. 4 ). Example: Kmeans clustering The Kmeans algorithm is used to ﬁnd 4 clusters in the Riply’s training set. data = load(’riply_trn’). figure. pboundary( model ).14).data.MsErr).’sk’. ppatterns(data). xlabel(’t’).is usually selected randomly. 6. The vectors classiﬁed to the corresponding centers are distinguished by color (see Figure 4. The plot of the objective function F (θ (t) ) with respect to the number of iterations is shown in Figure 4. % run kmeans % visualization figure. 3].6a).6b. plot(model. The Kmeans algorithm is implemented in the function kmeans. References: The Kmeans algorithm is described in the books [26. ylabel(’F’). 51 . hold on.y] = kmeans(data. ppatterns(model.X. The solution is visualized as the cluster centers.X.
5 1 (b) Figure 4.2 −1.5 0 0.5 −1 −0.2: Example of the MaximumLikelihood of the Gaussian Mixture Model from the incomplete data.2 0 −0.45 0.2 1 0.5 0 0.6 1.2 0 −0. 52 .5 1 (a) 1.5 −1 −0. Figure (a) shows components of the GMM and Figure (b) shows contour plot of the distribution function.2 1 0.6 0.4 0.981 0.8 gauss 2 0.2 −1.4 gauss 1 0.1.8 0.
Figure (a) shows ground truth and the estimated GMM and Figure(b) shows the loglikelihood with respect to the number of iterations of the EM.1 0.0. 53 .35 0.25 0.3: Example of using the EM algorithm for ML estimation of the Gaussian mixture model.15 0.2 0.3 0.05 0 −5 −4 −3 −2 −1 0 1 2 3 4 5 (a) −442 −444 −446 −448 log−likelihood −450 −452 −454 −456 −458 −460 0 5 10 iterations 15 20 25 (b) Figure 4.4 Ground truth ML estimation 0.
4 0.8 1 Figure 4.4 −0.4 gauss 1 0.4 −0.6 0.4: Example of the Minimax estimation of the Gaussian distribution.6 0. 54 .8 0.2 −0.2 0 −0.1 0.2 0 0.2 0.304 0.
2 0. 55 .1 0 −4 −3 −2 −1 0 1 svm output f(x) 2 3 4 5 (a) 1.7 0.5 1 (b) Figure 4. Figure (a) shows the sigmoid model and GMM model both ﬁtted by ML estimated.2 −1.5: Example of ﬁtting a posteriori probability to the SVM output.4 0.8 0.6 0.3 0. Figure (b) shows the corresponding SVM classiﬁer.5 0.6 p(y=1f(x)) 0.5 0 0.8 0.1 P(y=1f(x)) ML−Sigmoid P(y=1f(x)) ML−GMM P(f(x)y=1) ML−GMM P(f(x)y=2) ML−GMM 0.4 0.5 −1 −0.9 0.2 0 −0.2 1 0.
6 0.08 0.5 0 0.8 0.16 0. Figure (a) shows found clusters and Figure (b) contains the plot of the objective function F (θ (t) ) with respect to the number of iterations.14 0.2 1 0. 56 .2 −1.5 −1 −0.04 0 2 4 6 t 8 10 12 (b) Figure 4.1.1 F 0.4 0.2 0 −0.6: Example of using the Kmeans algorithm to cluster the Riply’s data set.06 0.5 1 (a) 0.12 0.
y ∈ Y = {1. . . . . (5. . αc ] be composed of all weight vectors and b = [b1 . si ∈ Rn which are usually subset of the training data. . . . s1 ).2) The datatype used by the STPRtool to represent the binary SVM classiﬁer is described in Table 5. 2.1 Support vector classiﬁers The binary support vector classiﬁer uses the discriminant function f : X ⊆ Rn → R of the following form f (x) = α · kS (x) + b . Both the binary and the multiclass SVM classiﬁers are implemented in the function svmclass.1) The kS (x) = [k(x. . y∈Y The datatype used by the STPRtool to represent the multiclass SVM classiﬁer is described in Table 5. c} is deﬁned as (5. 2} is deﬁned as q(x) = 1 for f (x) ≥ 0 . (5. . . . 2. sd }. k(x. . sd )]T is the vector of evaluations of kernel functions centered at the support vectors S = {s1 .Chapter 5 Support vector and other kernel machines 5. y∈Y. 57 . c} deﬁned as fy (x) = αy · kS (x) + by . . . The binary classiﬁcation rule q: X → Y = {1. Let the matrix A = [α1 . . . The multiclass classiﬁcation rule q: X → Y = {1. . .2. . The α ∈ Rl is a weight vector and b ∈ R is a bias.3. . . bc ]T be a vector of all biases.3) q(x) = argmax fy (x) . . . 2 for f (x) < 0 . The multiclass generalization involves a set of discriminant functions fy : X ⊆ Rn → R.
2 Binary Support Vector Machines The input is a set TXY = {(x1 . The jth rule qj (x) classiﬁes the inputs x into class yj ∈ Y or yj ∈ Y. i ∈ I1 ∪ I2 . . The strategy (5.b i=1 i ∈ I1 . . w · xi + b ≥ +1 − ξi . 2}. Let qj : X ⊆ Rn → {yj . b∗ ) is transformed to the following quadratic programming task l 1 ∗ ∗ 2 (w . The implemented methods to deal with the SVM and related kernel classiﬁers are enlisted in Table 5. The majority votes based multiclass classiﬁer assigns the input x into such class y ∈ Y having the majority of votes y = argmax vy (x) . g be a set of g binary 1 2 SVM rules (5. (xl . . .g The datatype used to represent the majority voting strategy for the multiclass SVM classiﬁer is described in Table 5. .4. . Let v(x) be a vector [c × 1] of votes for classes Y when the input x is to be classiﬁed. (5. (5. i ∈ I2 . yl )} of training vectors xi ∈ Rn and corresponding hidden states yi ∈ Y = {1. ∀y ∈ Y. vc (x)]T is computed as: Set vy = 0. g do vy (x) = vy (x) + 1 where y = qj (x). ξi ≥ 0 ... The training of the optimal parameters (w ∗ . . .5) 2 w. end..4) is implemented in the function mvsvmclass.2). b ) = argmin w + C ξip . j = 1. w · xi + b ≤ −1 + ξi . y1 ). 2 for f (x) < 0 . The vector v(x) = [v1 (x). .4) y =1. The linear SVM aims to train the linear discriminant function f (x) = w · x + b of the binary classiﬁer q(x) = 1 for f (x) ≥ 0 . 30] 5.. .1 References: Books describing Support Vector Machines are for example [31. For j = 1. . . .The majority voting strategy is other commonly used method to implement the 1 2 multiclass SVM classiﬁer. . . 4. . yj }. 58 .
59 . svmquadprog SVM trained by Matlab Optimization Toolbox. i ∈ I1 ∪ I2 are slack variables used to relax the inequalities for the case of nonseparable data. rbfpreimg2 RBF preimage problem by Gradient optimization. The linear p = 1 and quadratic p = 2 term for the slack variables ξi is used. The I1 = {i: yi = 1} and I2 = {i: yi = 2} are sets of indices. oaasvm Multiclass SVM using OneAgainstAll decomposition. smo Sequential Minimal Optimization for binary SVM with L1soft margin. kperceptr Kernel Perceptron. mvsvmclass Majority voting multiclass SVM classiﬁer. β ≤ 1C . bsvm2 Multiclass BSVM with L2soft margin. The ξi ≥ 0. evalsvm Trains and evaluates Support Vector Machines classiﬁer. kernel Evaluates kernel function. The C > 0 stands for the regularization constant.Table 5. minball Minimal enclosing ball in kernel feature space.6) subject to β·γ = 1. rsrbf Reduced Set Method for RBF kernel expansion. oaosvm Multiclass SVM using OneAgainstOne decomposition. The training of the nonlinear (kernel) SVM with L1soft margin corresponds to solving the following QP task β ∗ = argmax β · 1 − β 1 β · Hβ . svmlight Interface to SV M light software. kdist Computes distance between vectors in kernel space. svmclass Support Vector Machines Classiﬁer. rbfpreimg3 RBF preimage problem by KwokTsang’s algorithm. kfd Kernel Fisher Discriminant. In the case of the parameter p = 1.1: Implemented methods: Support vector and other kernel machines. diagker Returns diagonal of kernel matrix of given data. β ≥ 0. rbfpreimg RBF preimage by Schoelkopf’s ﬁxedpoint algorithm. 2 (5. demo svm Demo on Support Vector Machines. the L1soft margin SVM is trained and for p = 2 it is denoted as the L2soft margin SVM.
options. The discriminant function (5. . .Alpha [d × c] Weight vectors A = [α1 . . .arg [1 × p] Kernel argument(s). sd }. .fun = ’svmclass’ Identiﬁes function associated with this data type. . . . for i ∈ {j: yj = 2. xj ) if yi = yj . . The bias b is computed as an average over all the constrains such that 1 γi − α · kS (xi ) . Table 5. . . . . .3: Datatype used to describe multiclass SVM classiﬁer. The vector γ [l × 1] and the matrix H [l × l] are constructed as follows γi = 1 if yi = 1 . for i ∈ ISV . 0 < βj < C} . . must hold for the optimal solution.2: Datatype used to describe binary SVM classiﬁer. . xj ) if yi = yj .arg [1 × p] Kernel argument(s). The weight vectors are compute as αi = βi if yi = 1 . −βi if yi = 2 .fun = ’svmclass’ Identiﬁes function associated with this data type.1) is determined by the weight vector α. where k: X × X → R is selected kernel function.b [c × 1] Vector of biased b = [b1 .b [1 × 1] Bias b.options. Binary SVM classiﬁer (structure array): . αc ]. . −k(xi .ker [string] Kernel identiﬁer. . . where the set of indices is ISV = {i: βi > 0}. . bc ]T . . Multiclass SVM classiﬁer (structure array): . 0 < βj < C} .ker [string] Kernel identiﬁer. Hi. .options.sv. −1 if yi = 2 .options.Table 5. (5. . .X [n × d] Support vectors S = {s1 . . The set of support vectors is denoted as S = {xi : i ∈ ISV }.X [n × d] Support vectors S = {s1 .j = k(xi . b= Ib  i∈I b 60 . sd }.Alpha [d × 1] Weight vector α. bias b and the set of support vectors S. The 0 [l × 1] is vector of zeros and the 1 [l × 1] is vector of ones. .7) The bias b can be computed from the KarushKuhnTucker (KKT) conditions as the following constrains f (xi ) = α · kS (xi ) + b = +1 f (xi ) = α · kS (xi ) + b = −1 for i ∈ {j: yj = 1.sv.
. . .Alpha [d × g] Weight vectors A = [α1 .b [g × 1] Vector of biased b = [b1 . βj > 0} .bin y [2 × g] Targets for the binary rules. where E is the identity matrix. βj > 0} .4: Datatype used to describe the majority voting strategy for multiclass SVM classiﬁer. . The SV M light is described for instance in the book [27]. bg ]T . . . αg ].sv.:) 1 1 1 contains [y1 . . . . for i ∈ {i: yj = 2. . An interactive demo on the binary SVM is implemented in demo svm. The bias b is computed as an average over all the constrains thus 1 αi b= ) − α · kS (xi ) . γi (1 − ISV  i∈I 2C SV The binary SVM solvers implemented in the STPRtool are enlisted in Table 5.7). . 2 2C (5.8) subject to β·γ = 1. The weight vector α is given by (5. . The set of support vectors is set to S = {xi : i ∈ ISV } where the set of indices is ISV = {i: βi > 0}.fun = ’mvsvmclass’ Identiﬁes function associated with this data type. where Ib = {i: 0 < βi < C} are the indices of the vectors on the boundary.X [n × d] Support vectors S = {s1 . The bias b can be computed from the KKT conditions as the following constrains f (xi ) = α · kS (xi ) + b = +1 − αi 2C αi f (xi ) = α · kS (xi ) + b = −1 + 2C for i ∈ {i: yj = 1. . 61 . y2 . . . yg ]. sd }.:) contains 2 2 2 [y1 . yg ] and the bin y(2.arg [1 × p] Kernel argument(s). . 30]. 27. . β ≥ 0. . . The vector bin y(1. Example: Training binary SVM. . The training of the nonlinear (kernel) SVM with L2soft margin corresponds to solving the following QP task β ∗ = argmax β · 1 − β 1 1 β · (H + E)β . . . . Majority voting SVM classiﬁer (structure array): . References: The Sequential Minimal Optimizer (SMO) is described in the books [23.5.Table 5.ker [string] Kernel identiﬁer. .options. y2 . . must hold at the optimal solution. .options.
Function smo svmlight svmquadprog Table 5. ypred = svmclass(tst. options. cerror(ypred. It trains both L1soft and L2soft margin binary SVM. % train model = % model = % model = % % % % load training data use RBF kernel kernel argument regularization constant SVM classifier smo(trn. The decision boundary is visualized in Figure 5.1.tst. The binary SVM is trained on the Riply’s training set riply trn. svmlight(trn.5: Implemented binary SVM solvers. The SMO solver is used in the example but other solvers (svmlight.arg = 1. It can be downloaded from http://svmlight. It useful for small problems only. The executable ﬁle svm learn for the Linux OS must be installed.y) % load testing data % classify data % compute error 62 .model). options.X.options).00) which trains the binary SVM with L1soft margin. psvm(model). svmquadprog(trn.options). description Implementation of the Sequential Minimal Optimizer (SMO) used to train the binary SVM with L1soft margin.ker = ’rbf’. ppatterns(trn). SVM solver based on the QP solver quadprog of the Matlab optimization toolbox.org/. The function quadprog is a part of the Matlab Optimization toolbox.joachims. Matlab interface to SV M light software (Version: 5.options).C = 10. The trained SVM classiﬁer is evaluated on the testing data riply tst. % visualization figure. tst = load(’riply_tst’). trn = load(’riply_trn’). options. It can handle moderate size problems. svmquadprog) can by used as well. The SV M light is very powerful optimizer which can handle large problems.
5. yl )} of training vectors xi ∈ X ⊆ Rn and corresponding hidden states yi ∈ Y = {1. y∈Y. (xl . c}. . yl )} contain the modiﬁed hidden states deﬁned as 1 for y = yi . y1 ). . The OAA decomposition transforms the multiclass problem into a series of c binary subtasks that can be trained by the binary y SVM. The goal is to train the multiclass rule q: X ⊆ Rn → Y deﬁned by (5. yi = 2 for y = yi .8 0. .1: Example showing the binary SVM classiﬁer trained on the Riply’s data.0920 1. .2 1 0. y ∈ Y.3 OneAgainstAll decomposition for SVM The input is a set TXY = {(x1 .4 0. Let the training set TXY = {(x1 .5 −1 −0.5 1 Figure 5. . (xl . .ans = 0. The function allows to specify any binary SVM 63 .5 0 0.2 −1.3). .6 0. The discriminant functions fy (x) = αy · kS (x) + by . y1 ). 2. y are trained by the binary SVM solver from the set TXY . . The problem can be solved by the OneAgainstAll (OAA) decomposition. . . . The OAA decomposition is implemented in the function oaasvm.2 0 −0. .
X. . (xl .4). .arg = 1. options. Let the training set TXY = {(x1 . pboundary(model). 2 2 for yj = yi . ppatterns(model.solver (see Section 5. The SMO binary solver is used to train the binary SVM subtasks. % visualization figure.C = 10. . y1 ). j = 1. . The OAO decomposition for multiclass SVM is implemented in function 64 . The binary SVM rules qj . The OAO decomposition transforms the multiclass problem into a series of g = c(c − 1)/2 j binary subtasks that can be trained by the binary SVM.2. ppatterns(data). The goal is to train the multiclass rule q: X ⊆ Rn → Y based on the majority voting strategy deﬁned by (5. . Example: Multiclass SVM using the OAA decomposition The example shows the training of the multiclass SVM using the OAA decomposition.4 OneAgainstOne decomposition for SVM The input is a set TXY = {(x1 . The oneagainstone (OAO) decomposition strategy can be used train such a classiﬁer. ylj )} contain the training vectors xi ∈ I j = {i: yi = y 1 yi = y 2} and the modiﬁed the hidden states deﬁned as yi = 1 1 for yj = yi .3) represented by the datatype described in Table 5. . References: The OAA decomposition to train the multiclass SVM and comparison to other methods is described for instance in the paper [12]. The training data and the decision boundary is visualized in Figure 5. g is constructed for all g = c(c − 1)/2 combinations of 1 2 1 classes yj ∈ Y and yj ∈ Y \ {yj }. options. j The training set TXY . . 2. j = 1. . .3. options. model = oaasvm(data. .12).2) used to solve the binary problems.sv. . options. The result is the multiclass SVM classiﬁer (5. . c}. yl )} of training vectors xi ∈ X ⊆ Rn and corresponding hidden states yi ∈ Y = {1. . . . . .’ko’.ker = ’rbf’.options ). i ∈ Ij . g are trained on j the data TXY . . (xlj . . . % % % % % % load data use SMO solver use RBF kernel kernel argument regularization constant training 5.solver = ’smo’. data = load(’pentagon’). y1).
options.4 −0.4 0.8 −1 −1 −0.C = 10.3. % visualization figure.2 −0.ker = ’rbf’. oaosvm.sv.6 −0. options. The SMO binary solver is used to train the binary SVM subtasks. Example: Multiclass SVM using the OAO decomposition The example shows the training of the multiclass SVM using the OAO decomposition. pboundary(model).2 0 0. model = oaosvm(data.arg = 1.6 0. ppatterns(model.6 −0.2 0. ppatterns(data).options). options.8 −0.12).’ko’.6 0.2 0 −0.4 0.X.1 0. References: The OAO decomposition to train the multiclass SVM and comparison to other methods is described for instance in the paper [12]. The function oaosvm allows to specify an arbitrary binary SVM solver to train the subtasks.solver = ’smo’.2: Example showing the multiclass SVM classiﬁer build by oneagainstall decomposition.8 0. % % % % % % load data use SMO solver use RBF kernel kernel argument regularization constant training 65 .4 −0. data = load(’pentagon’). The training data and the decision boundary is visualized in Figure 5.8 1 Figure 5. options.
3: Example showing the multiclass SVM classiﬁer build by oneagainstone decomposition. y1 ). ξiy ≥ 0 . . ξ∗ ) = argmin ( wy 2 + b2 ) + C (ξi )p . c}. the optimization problem (5.8 −1 −1 −0.8 1 Figure 5.4 −0. . . ξ is a vector of slack variables and I = [1. .9) can be 66 . The letter B in BSVM stands for the bias term added to the objective (5.6 0. (xl . b∗ . l] is set of indices.6 0. 5.2 0. . bc ]T is vector of biases. In the case of the L2soft margin p = 2. . The goal is to train the multiclass rule q: X ⊆ Rn → Y deﬁned by (5.b. In the linear case. The L1soft margin is used for p = 1 and L2soft margin for p = 2.3).6 −0.4 −0. . . y ∈ Y \ {yi } . . .2 −0. the parameters of the multiclass rule can be found by solving the multiclass BSVM formulation which is deﬁned as 1 (W∗ .5 Multiclass BSVM formulation The input is a set TXY = {(x1 .2 0 −0.6 −0. . where W = [w 1 . yl )} of training vectors xi ∈ X ⊆ Rn and corresponding hidden states yi ∈ Y = {1. . 2. .ξ) subject to w yi · xi + byi − ( w y · xi + by ) ≥ 1 − ξiy .1 0.8 −0. .4 0. i ∈ I . b = [b1 . wc ] is a matrix of normal vectors.9).2 0 0.b i∈I y∈Y\{y} F (W. . (5.8 0. . .9) y 2 y∈Y W. i ∈ I .4 0. y ∈ Y \ {yi } . . . .
(5. yi) − δ(y.9) can be computed. . j) . The function h: Y×Y×I×I → R is deﬁned as h(y. Kozinec’s algorithm (solver = ’kozinec’). y)) + by . The bounds are used in the algorithms to deﬁne the following two stopping conditions: Relative tolerance: FU B − FLB ≤ εrel . The multiclass rule (5. y ∈ Y is given by by = i∈I y∈Y\{yi } u αi (δ(y.10) subject to y αi ≥ 0 . u. i ∈ I . yi) − δ(u. j) = ( xi ·xj +1)(δ(yi . .10) can be solved by most the optimizers. u)−δ(yi .10) corresponds to the singleclass SVM criterion which is simple to optimize.12) where the bias by . yj )+δ(y. ξ ∗ ) of the criterion (5. The upper bound FU B and the lower bound FLB on the optimal value F (W∗. where A = [α1 . The optimizers implemented in the function bsvm2 are enlisted bellow: MitchellDemyanovMalozemov (solver = ’mdm’). j) . y∈Y. u)) . y))+ δ(y.12).expressed in its dual form A∗ = argmax A i∈I y∈Y\{yi } y αi − 1 2 y u αi αj h(y. The nonlinear (kernel) classiﬁer is obtained by substituting the selected kernel k: X × X → R for the dot products x·x to (5. FLB + 1 67 . i∈I y∈Y\{yi } j∈I u∈Y\{yj } (5. The above mentioned algorithms are of iterative nature and converge to the optimal solution of the criterion (5. . b∗ . αc ] are optimized weight vectors. . y∈Y. u)δ(i. Nearest Point Algorithm (solver = ’npa’).3) is composed of the set of discriminant functions fy : X → R which are computed as fy (x) = i∈I xi · x u∈Y\{yi } u αi (δ(u.11) and (5. (5. i.9). u)−δ(yj . i. The optimization task (5. u. y ∈ Y \ {yi} . The training of the multiclass BSVM classiﬁer with L2soft margin is implemented in the function bsvm2.11) 2C The criterion (5.
.3). y1). . % % % % % % load data use NPA solver use RBF kernel kernel argument regularization constant training 5. The input training vectors are assumed to be mapped into a new feature space F by a mapping function φ: X → F . (φ(xl ). The training data and the decision boundary is visualized in Figure 5. yl )}. . options. α · (N + µE)α α · (N + µE)α 68 (5. Example: Multiclass BSVM with L2soft margin. pboundary(model).6 Kernel Fisher Discriminant The input is a set TXY = {(x1 . The MitchellDemyanovMalozemov and Nearest Point Algorithm are described in paper [14].’ko’. . options.12). 2}. The Kernel Fisher Discriminant is the nonlinear extension of the linear FLD (see Section 2. . The mapping is performed eﬃciently by means of kernel functions k: X × X → R being the dot products of the mapped vectors k(x. . .X. .options). (xl . yl )} of training vectors xi ∈ X ⊆ Rn and corresponding values of binary state yi ∈ Y = {1. model = bsvm2(data. % visualization figure. αl ]T such that the criterion F (α) = α·m 2 α · Mα = . .arg = 1.4. ppatterns(data). The aim is to ﬁnd a direction ψ = l αi φ(xi ) in the feature space i=1 F given by weights α = [α1 . data = load(’pentagon’).ker = ’rbf’. The Kozinec’s algorithm based SVM solver is described in paper [10]. The transformation of the multiclass BSVM criterion to the singleclass criterion (5. options. ’mdm’).C = 10.solver = ’npa’. The example shows the training of the multiclass BSVM using the NPA solver. options. The other solvers can be used instead by setting the variable options.sv. .13) . The ordinary FLD is applied on the mapped data training data TΦY = {(φ(x1 ). x ) = φ(x) · φ(x ) .10) implemented in the STPRtool is published in paper [9]. y1 ).solver (other options are ’kozinec’.Absolute tolerance: FU B ≤ εabs . . References: The multiclass BSVM formulation is described in paper [12]. ppatterns(model. .
8 1 Figure 5. . The criterion (5. m = m1 − m2 . . The kernel matrix K [l × l] contains kernel evaluations Ki.4 −0. .8 −0. y where the vector 1y has entries i ∈ Iy equal to and remaining entries zeros.8 0. Iy  N = KKT − y∈Y y∈Y.4 0.14) where the k(x) = [k(x1 . .6 0.j = k(xi . The diagonal matrix µE in the denominator of the criterion (5. (5. matrices N [l × l] M [l × l] are constructed as my = 1 K1y .13) is the feature space counterpart of the criterion (2. k(xl .14) is known up to the bias b which is determined by the linear SVM with L1soft margin is applied on the projected data TZY = {(z1 . 69 . Iy my mT . The discriminant function f (x) of the binary classiﬁer (5.13) deﬁned in the input space X . The vector m [l × 1]. . The projections {z1 .2 0 −0. (zl .6 0. x). .2 −0.1 0.4 −0.8 −1 −1 −0. . . .6 −0.6 −0.4: Example showing the multiclass BSVM classiﬁer with L2soft margin. x)]T is vector of evaluations of the kernel. . . xj ).2 0.2) is given by the found vector α as f (x) = ψ · φ(x) + b = α · k(x) + b . The KFD training algorithm is implemented in the function kfd. y1). . The discriminant function (5.4 0. is maximized. M = (m1 − m2 )(m1 − m2 )T . yl )}.13) serves as the regularization term. zl } are computed as zi = α · k(xi ) .2 0 0.
X.15) . options.C = 10. % evaluation on testing data tst = load(’riply_tst’). ppatterns(trn).5. cerror( ypred.y ) ans = 0.ker = ’rbf’. The classiﬁer is visualized as the SVM classiﬁer. xb ) = φ(xa ) · φ(xb ) . The mapping is performed implicitly by using the kernel functions k: X × X → R which are dot products of the input data mapped into the feature space F such that k(xa .arg = 1. options.0960 % % % % % % load training data use RBF kernel kernel argument regularization of linear SVM regularization of KFD criterion KFD training 5. . i. options. model ).mat.e.. tst. . The kernel methods works with the data nonlinearly mapped φ: X → F into a new feature space F . model = kfd(trn. the isolines with f (x) = +1 and f (x) = −1 are visualized and the support vectors (whole training set) are marked as well. xl } be a set of vector from the input space X ⊆ Rn .7 Preimage problem Let TX = {x1 .References: The KFD formulation is described in paper [19]. The detailed analysis can be found in book [30]. . The training data and the found kernel classiﬁer is visualized in Figure 5. . Example: Kernel Fisher Discriminant The binary classiﬁer is trained by the KFD on the Riply’s training set riply trn. The found classiﬁer is evaluated on the testing data riply tst. psvm(model). Let ψ ∈ F be given by the following kernel expansion l ψ= i=1 αi φ(xi ) .mu = 0. options) % visualization figure. options.mat. trn = load(’riply_trn’). 70 (5. ypred = svmclass( tst.001.
its feature space image φ(x) ∈ F would well approximate the expansion point Ψ ∈ F given by (5.2 −1.2 0 −0.16) for the particular case of the RBF kernel leads to the following task l x = argmax x i=1 αi exp(−0. Let the Radial Basis Function (RBF) k(xa .2 1 0. the preimage problem solves mapping from the feature space F back to the input space X . all the methods are guaranteed to ﬁnd only a local optimum of the task (5.5 0 0.5 x − xi 2 /σ 2 ) . xb ) = exp(−0. (5.8 0. The preimage problem (5. .5 xa −xb 2 /σ 2 ) be assumed. The preimage problem can be stated as the following optimization task x = argmin φ(x) − ψ x l 2 = argmin φ(x ) − x i=1 αi φ(xi ) l 2 (5. . The preimage problem aims to ﬁnd an input vector x ∈ X .6 0. However. .17).5 −1 −0.16) l l = argmin k(x . x ) − 2 x i=1 αi k(x . xj ) .5 1 Figure 5.4 0. 71 . where α = [α1 . In other words.1. .17) which are enlisted in Table 5. αl ]T is a weight vector of real coeﬃcients.15).5: Example shows the binary classiﬁer trained based on the Kernel Fisher Discriminant.6. xi ) + i=1 j=1 αi αj k(xi .17) The STPRtool provides three optimization algorithms to solve the RBF preimage problem (5.
1)/num_data.13).options. expansion. References: The preimage problem and the implemented algorithms for its solution are described in [28.arg = 1. % define the mean in the feature space num_data = size(data. % visualization figure. expansion. expansion.X = data.Table 5.’or’. % load input data data = load(’riply_trn’). ppatterns(x. Example: Preimage problem The example shows how to visualize a preimage of the data mean in the feature induced by the RBF kernel.6. rbfpreimg3 A method exploiting the correspondence between distances in the input space and the feature space proposed in [16]. rbfpreimg An iterative ﬁxed point algorithm proposed in [28].X).X.ker = ’rbf’.2). expansion.sv. 72 . ppatterns(data. 16]. % compute preimage problem x = rbfpreimg(expansion). The result is visualized in Figure 5.options.6: Methods to solve the RBF preimage problem implemented in the STPRtool. rbfpreimg2 A standard gradient optimization method using the Matlab optimization toolbox for 1D search along the gradient.Alpha = ones(num_data.
Let φ: X → F be a mapping from the input space X to the feature space F associated with kernel function such that k(xa . . αl ]T is a weight vector of real coeﬃcients and b ∈ R is a scalar. m < l.8 Reduced set method The reduced set method is useful to simplify the kernel classiﬁer trained by the SVM or other kernel machines.4 0. . si ∈ Rn determine the reduced kernel expansion (5. (5.2 −1. .19) is shorter. The weight vector β = [β1 .5 −1 −0. xl } are vectors from Rn . . .8 0. . The function f (x) can be expressed in the feature space F as the linear function f (x) = ψ · φ(x) + b. and well approximates the original one (5.5 0 0.19). . i. .6: Example shows training data and the preimage of their mean in the feature space induced by the RBF kernel. .18). .2 1 0. The reduced set method aims to ﬁnd a new function m m ˜ f (x) = i=1 βi k(si . The problem of ﬁnding the 73 .18) where TX = {x1 . sm }. x) + b = ψ · φ(x) + b .6 0. βm ]T and the vectors TS = {s1 . x) + b = φ(x) · i=1 βi φ(si ) + b . . . . . 5. α = [α1 .19) such that the expansion (5. ˜ ψ (5. . k: X × X → R is a kernel function. Let the function f : X ⊆ Rn → R be given as l f (x) = i=1 αi k(xi .5 1 Figure 5.. . xb ) = φ(xa ) · φ(xb ) .2 0 −0.1.e.
struct(’nsv’. The algorithm converts the task (5. % compute the reduced rule with 10 support vectors red_model = rsrbf(model.10)).y).reduced kernel expansion (5.struct(’line_style’. ypred = svmclass(tst.struct(’line_style’.’r’)). The reduced set method for the RBF kernel is implemented in function rsrbf. References: The implemented reduced set method is described in [28]. red_ypred = svmclass(tst. ppatterns(trn). TS ) = argmin ψ − ψ β .’C’.20) into a series of m preimage tasks (see Section 5.1. In this case.7).’b’)).10)).19) can be stated as the optimization task m l 2 ˜ (β.’rbf’. xb ) = exp(−0. 74 . the optimization task (5. The trained kernel expansion (discriminant function) is composed of 94 support vectors.’arg’.mat.model). h2 = pboundary(red_model.X.7. The original and reduced SVM are evaluated on the testing data riply tst.struct(’ker’.20) Let the Radial Basis Function (RBF) k(xa .20) can be solved by an iterative greedy algorithm.TS 2 = argmin β . (5.’Original SVM’. err = cerror(ypred. Decision boundaries of the original and the reduced SVM are visualized in Figure 5.5 xa −xb 2 /σ 2 ) be assumed.tst. legend([h1(1) h2(1)]. The reduced expansion with 10 support vectors is found. % train SVM classifier model = smo(trn. % load training data trn = load(’riply_trn’). % visualize decision boundaries figure. h1 = pboundary(model.’Reduced SVM’). The SVM classiﬁer is trained from the Riply’s training set riply trn/mat.model). % evaluate SVM classifiers tst = load(’riply_tst’).TS i=1 βi φ(si ) − i=1 αi φ(xi ) .X. Example: Reduced set method for SVM classiﬁer The example shows how to use the reduced set method to simplify the SVM classiﬁer.
2 Original SVM Reduced SVM 1 0.y).nsv.2 −1. The original decision rule involves 94 support vectors while the reduced one only 10 support vectors. err = %.2 0 −0.. err. err = 0.nsv. ’Reduced SVM: nsv = %d.5 −1 −0. 75 . model.tst.5 1 Figure 5. Original SVM: nsv = 94. err = 0.6 0. red_err).7: Example shows the decision boundaries of the SVM classiﬁer trained by SMO and the approximated SVM classiﬁer found by the reduced set method..8 0.red_err = cerror(red_ypred.4f\n’ . fprintf([’Original SVM: nsv = %d.0960 1.4f\n’]. err = %.4 0. red_model.0960 Reduced SVM: nsv = 10...5 0 0. .
mat /svm samples/*. Training part of the Riply’s data set [24]. Input data for the demo on the ExpectationMaximization algorithm for the Gaussian mixture model demo emgmm. The datasets stored in the Matlab data ﬁle are enlisted in Table 6.mat 76 .mat /riply tst. /multi separable/*.1: Datasets /riply trn. Table 6.1 Data sets The STPRtool provides several datasets and functions to generate synthetic data.mat stored in /data/ directory of the STPRtool.Chapter 6 Miscellaneous 6.1.mat /gmm samles/*. Well known Fisher’s data set. The functions which generate synthetic data and scripts for converting external datasets to the Matlab are given in Table 6.mat /iris. Testing part of the Riply’s data set [24]. Examples of handwritten numerals.mat /binary separable/*.1. Input data for the demo on the Generalized Anderson’s task demo anderson. Input data for the demo on the Support Vector Machines demo svm.mat /anderson task/*.mat /ocr data/*. A description is given in Section 7. Linearly separable multiclass data in 2D.2. Input data for the demo on training linear classiﬁers from ﬁnite vector sets demo linclass.
pboundary Plots decision boundary of given classiﬁer in 2D. Postal Service (USPS) database of handwritten numerals to the Matlab ﬁle. showim Displays images entered as column vectors. Generates data from the Gaussian mixture model. GUI which allows to create interactively labeled vectors sets and labeled sets of Gaussian distributions. 2D and 3D. Each function contains an example of using. plane3 Plots plane in 3D. Table 6. pandr 77 . pgmm Visualizes the bivariate Gaussian mixture model. pline Plots line in 2D.createdata genlsdata gmmsamp gsamp gencircledata usps2mat Table 6. ppatterns Visualizes patterns as points in the feature space. Generates data from the multivariate Gaussian distribution. Generates linearly separable binary data with prescribed margin. It works for 1D. The list of functions for visualization is given in Table 6.S.2 Visualization The STPRtool provides tools for visualization of common models and data.3.2: Data generation. It operates in 2dimensional space only. 6. Visualizes solution of the Generalized Anderson’s task.3: Functions for visualization. psvm Plots decision boundary of binary SVM classiﬁer in 2D. pkernelproj Plots isolines of kernel projection. Generates data on circle corrupted by the Gaussian noise. pgauss Visualizes set of bivariate Gaussians. Converts U.
. (6.1) with the 0/1loss function corresponds to the expectation of misclassiﬁcation.6.1) is referred to as the Bayesian rule q ∗ (x) = argmin y y∈Y PXY (x. A decision rule q: X → D takes a decision d ∈ D based on the observation x ∈ X . y)W (q(x). 1 for q(x) = y . 1 for q(x) = y . .3) = argmax PXY (xy)PY (y) . y) . The loss function is deﬁned as ⎧ ⎨ 0 for q(x) = y .3 Bayesian classiﬁer The object under study is assumed to be described by a vector of observations x ∈ X and a hidden state y ∈ Y. y) = 0 for q(x) = y . R(q) = X y∈Y PXY (x. y). (6. y∈Y 78 . c}. The rule q: X → Y which minimizes the Bayesian risk with the loss function (6. y)W (q(x). Let W : D × Y → R be a loss function which penalizes the decision q(x) ∈ D when the true hidden state is y ∈ Y. (6. The x and y are realizations of random variables with joint probability distribution PXY (x. The Bayesian risk R(q) is an expectation of the value of the loss function W when the decision rule q is applied. y) dx .5) q(x) = ⎩ dont know if 1 − max PY X (yx) ≥ ε . . The Bayesian risk (6.e.. y∈Y (6. . y∈Y Classiﬁcation with rejectoption: The set of decisions D is assumed to be D = Y ∪ {dont know}. The rule q: X → Y which minimizes the expectation of misclassiﬁcation is deﬁned as q(x) = argmax PY X (yx) . The STPRtool implements the Bayesian rule for two particular cases: Minimization of misclassiﬁcation: The set of decisions D coincides to the set of hidden states Y = {1. where ε is penalty for the decision dont know.1) The optimal rule q ∗ which minimizes the Bayesian risk (6. y∈Y y∈Y (6. y) = ⎩ ε for q(x) = dont know .4) is deﬁned as ⎧ ⎨ argmax PXY (xy)PY (y) if 1 − max PY X (yx) < ε .2) is used. i.4) Wε (q(x). Let X ⊆ Rn and the sets Y and D be ﬁnite. ∀x ∈ X . The 0/1loss function W0/1 (q(x).
The classconditional distributions PXY are modeled by the Gaussian mixture models (GMM) with two components.inx1). .4: Datatype used to describe the Bayesian classiﬁer. References: The Bayesian decision making with applications to PR is treated in details in book [26].struct(’ncomp’. The designed Bayesian classiﬁer is evaluated on the testing data riply tst. 2} are estimated by the relative occurrences (it corresponds to the ML estimate). The Bayesian classiﬁers (6. Table 6.X(:. % Evaluation on testing data tst = load(’riply_tst’).X(:.2. . % Estimation of priors n1 = length(inx1).Pclass{2} = emgmm(trn.Pclass{1} = emgmm(trn.mat.mat. y ∈ Y.tst. inx1 = find(trn.5) are implemented in function bayescls. % Estimation of classconditional distributions by EM bayes_model.2)).X. cerror(ypred. Bayesian classiﬁer (structure array): .fun = ’bayescls’ Identiﬁes function associated with this data type. y ∈ {1.To apply the optimal classiﬁcation rules one has to know the classconditional distributions PXY and a priory distribution PY (or their estimates). ypred = bayescls(tst. The distribution PXY (xy) is given by Pclassy which uses the datatype described in Table 4.3) and (6. The a priori probabilities PY (y).Prior = [n1 n2]/(n1+n2). n2 = length(inx2).4. % load input training data trn = load(’riply_trn’). Example: Bayesian classiﬁer The example shows how to design the plugin Bayesian classiﬁer for the Riply’s training set riply trn. bayes_model.struct(’ncomp’. bayes_model.2)). The datatype used by the STPRtool is described in Table 6.y==2).inx2). inx2 = find(trn.Pclass [cell 1 × c] Class conditional distributions PXY .bayes_model).Prior [1 × c] Discrete distribution PY (y).y) 79 . The GMM are estimated by the EM algorithm emgmm.y==1).
0900 The following example shows how to visualize the decision boundary of the found Bayesian classiﬁer minimizing the misclassiﬁcation (solid black line). % Vislualization of rejetoption rule pboundary(reject_model. % Visualization figure. % Penalization for don’t know decision reject_model = bayes_model. hold on. the second class and the region in between corresponding to the dont know decision.1.fun = ’bayescls’.4 Quadratic classiﬁer The quadratic classiﬁcation rule q: X ⊆ Rn → Y = {1.struct(’line_style’. . 2. reject_model. 80 .eps = 0. .ans = 0. 6. The Bayesian classiﬁer splits the feature space into two regions corresponding to the ﬁrst class and the second class. The decision boundary of the rejectoptions rule (dashed line) is displayed for comparison. The visualization is given in Figure 6. ∀y ∈ Y . c} is composed of a set of discriminant functions fy (x) = x · Ay x + by · x + cy . a vector by [n × 1] and a scalar cy [1 × 1]. . The input vector x ∈ Rn is assigned to the class y ∈ Y its corresponding discriminant function fy attains maximal value y = argmax fy (x) = argmax ( x · Ay x + by · x + cy ) . pboundary(bayes_model). . ppatterns(trn). the quadratic classiﬁer is represented by a single discriminant function f (x) = x · Ax + b · x + c . The rejectoption rule splits the feature space into three regions corresponding to the ﬁrst class.’k’)). The quadratic discriminant function fy is determined by a matrix Ay [n × n]. y ∈Y y ∈Y (6.1. which are quadratic with respect to the input vector x ∈ Rn .6) In the particular binary case Y = {1. bayes_model. 2}.
a vector b [n × 1] and a scalar c [1 × 1]. Binary linear quadratic (structure array): . The classiﬁer is trained from the Riply’s 81 .7) The STPRtool uses a speciﬁc structure array to describe the binary (see Table 6.fun = ’quadclass’ Identiﬁes function associated with this data type. given by a matrix A [n × n].2 1 0.5 1 Figure 6.5: Datatype used to describe binary quadratic classiﬁer.5 −1 −0. Example: Quadratic classiﬁer The example shows how to train the quadratic classiﬁer using the Fisher Linear Discriminant and the quadratic data mapping. 2 if f (x) = x · Ax + b · x + c < 0 . Table 6.A [n × n] Parameter matrix A. The classconditional probabilities are modeled by the Gaussian mixture models with two components estimated by the EM algorithm.4 0. . The implementation of the quadratic classiﬁer itself provides function quadclass.b [n × 1] Parameter vector b.8 0.1 (dashed line).5 0 0. .6) quadratic classiﬁer. 2} as follows q(x) = 1 if f (x) = x · Ax + b · x + c ≥ 0 .2 −1.2 0 −0.5) and the multiclass (see Table 6.6 0. . The input vector x ∈ Rn is assigned to class y ∈ {1.1: Figure shows the decision boundary of the Bayesian classiﬁer (solid line) and the decision boundary of the rejectoption rule with ε = 0. (6.1.c [1 × 1] Parameter scalar c.
(xl .5 Knearest neighbors classiﬁer Let TXY = {(x1 . % make quadratic classifier quad_model = lin2quad(lin_model). . . y ∈ Y. cerror(ypred. ppatterns(trn). The Knearest neighbor (KNN) classiﬁcation rule q: X → Y is deﬁned as q(x) = argmax v(x.y) ans = 0. . The found quadratic classiﬁer is evaluated on the testing data riply trn. l}.mat. .fun = ’quadclass’ Identiﬁes function associated with this data type. ypred = quadclass(tst. .Table 6.tst. i ∈ {1. {xi : xi ∈ Rn (x)} = K. y1 ). . c}. i.8) y∈Y 82 .. Multiclass quadratic classiﬁer (structure array): .e. trn = load(’riply_trn’).6: Datatype used to describe multiclass quadratic classiﬁer.0940 % load input data % quadratic mapping % train FLD 6. y ∈ Y.quad_model). . . . Let Rn (x) = {x : x − x ≤ r 2 } be a ball centered in the vector x in which lie K prototype vectors xi . . y ∈ Y. .2). lin_model = fld(quad_data).X. quad_data = qmap(trn). y) .b [n × c] Parameter vectors by . yl )} be a set of prototype vectors xi ∈ X ⊆ Rn and corresponding hidden states yi ∈ Y = {1. . % visualization figure. training set riply trn. .c [1 × c] Parameter scalars cy . The boundary of the quadratic classiﬁer is visualized (see Figure 6. .A [n × n × c] Parameter matrices Ay .mat. (6. pboundary(quad_model). % evaluation tst = load(’riply_tst’). .
% load training data and setup 8NN rule trn = load(’riply_trn’).X [n × l] Prototype vectors {x1 . .1. .6 0. Example: Knearest neighbor classiﬁer The example shows how to create the (K = 8)NN classiﬁcation rule from the Riply’s training data riply trn.4 0.3. yl }.8 0. . . The classiﬁer is evaluated on the testing data riply tst. 83 . The classiﬁcation rule (6. in book [6].mat. .8). for instance. where v(x.7: Datatype used to describe the Knearest neighbor classiﬁer. Table 6. . The decision boundary is visualized in Figure 6.5 1 Figure 6.2 0 −0. KNN classiﬁer (structure array): . xl }.5 −1 −0.7.2: Figure shows the decision boundary of the quadratic classiﬁer. The datatype used by the STPRtool to represent the Knearest neighbor classiﬁer is described in Table 6. References: The Knearest neighbors classiﬁer is described in most books on PR.fun = ’knnclass’ Identiﬁes function associated with this data type.y [1 × l] Labels {y1 .5 0 0. y) is number of prototype vectors xi with hidden state yi = y which lie in the ball xi ∈ Rn (x).2 1 0. .2 −1. model = knnrule(trn. .mat.8) is computationally demanding if the set of prototype vectors is large. . .
% evaluate classifier tst = load(’riply_tst’).3: Figure shows the decision boundary of the (K = 8)nearest neighbor classiﬁer.8 0. 84 .5 0 0. ypred = knnclass(tst.tst.2 1 0.6 0.model).5 −1 −0.% visualize decision boundary and training data figure.1160 1. pboundary(model).5 1 Figure 6. cerror(ypred.4 0.y) ans = 0. ppatterns(trn).2 0 −0.X.2 −1.
Chapter 7 Examples of applications We intend to demonstrate how methods implemented in the STPRtool can be used to solve a more complex applications. The function save chars saves drawn numerals in the stage of collecting training examples and the function ocr fun is used to recognize the numerals if the trained OCR is applied.1a). The STPRtool contains examples of numerals collected 85 . the functions save chars and ocr fun are called from the mpaper to process drawn numerals. The STPRtool provides GUI which allows to use standard computer mouse as an input device to draw images of numerals. The function collect chars invokes the form for drawing numerals and allows to save the numerals to a speciﬁed ﬁle. In particular. The training stage of the OCR system involves (i) collection of training examples of numerals and (ii) training the multiclass SVM.2). The GUI is intended just for demonstration purposes and the scanner or other device would be used in practice instead. The function mpaper creates a form composed of 50 cells to which the numerals can be drawn (see Figure 7.1) and image denoising system using the kernel PCA (Section 7. The GUI for drawing numerals is implemented in a function mpaper. The set of training examples has to be collected before the OCR is trained. The applications selected are the optical character recognition (OCR) system based on the multiclass SVM (Section 7.1 Optical Character Recognition system The use of the STPRtool is demonstrated on a simple Optical Character Recognition (OCR) system for handwritten numerals from 0 to 9. 7. The drawn numerals are normalized to size 16 × 16 pixels (diﬀerent size can be speciﬁed). A numeral to be classiﬁed is represented as a vector x containing pixel values taken from the numeral image. The function mpaper is designed to invoke a speciﬁed function whenever the middle mouse button is pressed. The multiclass SVM are used as the base classiﬁer of the numerals. To see the trained OCR system run the demo script demo ocr.
if a speed of classiﬁcation is the major objective the multiclass BSVM formulation is often a better option as it produce less number of support vectors. σ1 ). yl )} created the multiclass SVM classiﬁer can be trained. The directory /demo/ocr/models/ contains the SVM model trained by OAO decomposition strategy (function oaosvm) and model trained based on the multiclass BSVM formulation (function bsvm2).mat (the character can be omitted from the name) is used. σk )}. . The string Directory speciﬁes the directory where the ﬁles name xx.from 6 diﬀerent persons. . However. The training procedure for OCR classiﬁer is implemented in the function train ocr. xb ) with its argument. y1 ). 86 . The parameter tuning is implemented in the script tune ocr. . . After the parameters (C ∗ . The SVM has two free parameters which have to be tuned: the regularization constant C and the kernel function k(xa . σ ∗ ) are tuned the SVM classiﬁer is trained using all training data available. σ ∗ ) turned out to be have the lowest crossvalidation error over the given set Θ.mat are stored and the string FileName is a ﬁle name to which the resulting training set is stored. . The resulting ﬁle contains a matrix X of all examples and a vector y containing labels are assigned to examples according the number xx. In particular. the OneAgainstOne (OAO) decomposition was used as it requires the shortest training time.5 xa − xb 2 /σ 2 ) determined by the parameter σ was used in the experiment. the free SVM parameters are C and σ. If the ﬁle name format name xx. (Ck . . The methods to train the multiclass SVM are described in Chapter 5. (xl . Figure 7.DataFile) can be used to merge all the ﬁle to a single one. The function evalsvm returns the SVM model its parameters (C ∗ .1a shows the form ﬁlled with the examples of the numeral ’1’. A common method to tune the free parameters is to use the crossvalidation to select the best parameters from a preselected set Θ = {(C1 . The string name is name of the person who drew the particular numeral and xx stands for a label assigned to the numeral. . Therefore. For example. The vector dimension n = 256 corresponds to 16 × 16 pixels of each image numeral and l = 50 is the maximal number of numerals drawn to a single form. The RBF kernel k(xa . The images of normalized numerals are stored as vectors to a matrix X of size [256 × 50]. xb ) = exp(−0. The ﬁle name format name xx. The labels from y = 1 to y = 9 corresponds to the numerals from ’1’ to ’9’ and the label y = 10 is used for numeral ’0’. .mat is met then the function mergesets(Directory.1b shows numerals after normalization which are saved to the speciﬁed ﬁle honza cech.mat. The numerals are stored in the directory /data/ocr numerals/. The STPRtool contains a function evalsvm which evaluates the set of SVM parameters Θ based on the crossvalidation estimation of the classiﬁcation error. The Figure 7. Each person drew 50 examples of each of numerals from ’0’ to ’9’. The parameters C = ∞ (data are separable) and σ = 5 were found to be the best. Having the labeled training set TXY = {(x1 . to collect examples of numeral ’1’ from person honza cech collect_chars(’honza_cech1’) was called.
A link between the input space X and the feature space F is given by a mapping φ: X → F which is implicitly determined by a selected kernel function k: X × X → R. . The kernel PCA was used to model the image class. The model should describe the nonlinear subspace Ximg ⊆ Rn where the images live. A noisy image x is assumed to lie outside the subspace Ximg .2. It is implemented in the function usps2mat.2 Image denoising The aim is to train a model of a given class of images in order to restore images corrupted by noise. k(xa . The subspace Fimg is modeled by the kernel PCA trained from the set TX mapped into the feature space F by a function φ: X → F . Thus the unlabeled training data set TX = {x1 . . It is assumed that the nonlinear subspace Ximg forms a linear subspace Fimg of a new feature space F . The U.mat its contains is summarized in Table 7. . Having the kernel PCA model of the subspace Fimg it can be ˆ used to map the images x ∈ X into the subspace Ximg .e. xl } is used. The kernel function corresponds to the dot product in the feature space. . φ(xl )} denote the feature space representation of the training images TX . The STPRtool contains a script which converts the USPS database to the Matlab data ﬁle.The particular model used by OCR is speciﬁed in the function ocr fun which implicitly uses the ﬁle ocrmodel. The kernel PCA can be used to ﬁnd the basis vectors of the subspace Fimg from the input training data TX . To run the resulting OCR type demo ocr or call mpaper(struct(’fun’.3.S. The kernel PCA computes the lower dimensional representation TZ = {z 1 . Let TΦ = {φ(x1 ).. . . Postal Service (USPS) database of images of handwritten numerals [17] was used as the class of images to model. . The idea of using the kernel PCA for image denoising is demonstrated in Figure 7.’ocr_fun’)) If one intends to modify the classiﬁer or the way how the classiﬁcation results are displayed then modify the function ocr fun. The denoising procedure is based on mapping the noisy ˆ image x into the subspace Ximg .mat stored in the directory /demo/ocr/. . .1. The corruption of the USPS images by the Gaussian noise is implemented in script make noisy data. xb ) = φ(xa ) · φ(xb ) . z l }. This method was proposed in [20] and further developed in [15]. 7. . i. . z i ∈ Rm of the mapped images 87 . The aim is to train a model of images of all numerals at ones. A snapshot of running OCR is shown in Figure 7. The example which produced the ﬁgure is implemented in demo kpcadenois. The additive Gaussian noise was assumed as the source of image degradation. The script make noisy data produces the Matlab ﬁle usps noisy. .
the nonlinearity hidden in 88 .1: The content of the ﬁle usps noisy. This procedure is implemented in function kpcarec. .2) which yields the resulting x . .2) The Radial Basis Function (RBF) kernel k(xa . . (7.5 xa − xb 2 /σ 2 ) was used here as the preimage problem (7. X [256 × 2007] Training images corrupted by the Gaussian noise with signaltonoise ration SNR = 1 (it can be changed). the kernel PCA allows to model the nonlinearity which is very likely in the case of images. x)]T contains kernel functions evaluated at the training data TX . z = AT k(x) + b . . .1) The vector k(x) = [k(x1 . y [1 × 2007] Labels y ∈ Y = {1. . . . .2) can be well solved for this case. i=1 β = A(z − b) . (7. Structure array tst containing the testing data: gnd X [256 × 2007] Testing images of numerals represented as column vectors. 10} of training images. In contrast to the linear PCA.mat produced by the script make noisy data. k(xl .Table 7. . The input are 2007 grayscale images of size 16 × 16. . The matrix A [l × m] and the vector b [m × 1] are trained by the ˜ kernel PCA algorithm. xb ) = exp(−0. 10} of training images. Given a projection φ(x) onto the kernel PCA subspace an n image x ∈ R of the input space can be recovered by solving the preimage problem ˜ x = argmin φ(x) − φ(x) x 2 . TΦ to minimize the mean square reconstruction error l εKM S = i=1 ˜ φ(xi ) − φ(xi ) 2 . . X [256 × 7291] Training images corrupted by the Gaussian noise with signaltonoise ration SNR = 1 (it can be changed).1) to ˜ compute z which determines φ(x) and (ii) solving the preimage problem (7. y [1 × 7291] Labels y ∈ Y = {1. where ˜ φ(x) = l βi φ(x) . x). Structure array trn containing the training data: gnd X [256 × 7291] Training images of numerals represented as column vectors. The whole ˆ procedure of reconstructing the corrupted image x ∈ Rn involves (i) using (7. The input are 7291 grayscale images of size 16 × 16. However.
the function pcarec was used.the parameter σ of the used RBF kernel has to be tuned properly to prevent overﬁtting. The whole training procedure of the kernel PCA is implemented in the script train kpca denois. The results of tuning the parameters are displayed in Figure 7.1) and (7. In the case of the linear PCA.4 shows the idea of tuning the parameters of the kernel PCA. It produces sparse model and can deal with large data sets. m) = X was minimal. m1 ). The linear PCA was used to solve the same task for comparison. Figure 7.5. mk )} such that the crossvalidation estimate of the mean square error PX (x) x − x 2 dx . (σk . . The same idea was used to tune the output dimension of the linear PCA. Another parameter is the dimension m of the subspace kernel PCA subspace to be trained. . The images are assumed to be generated from distribution PX (x).2). The ﬁgure was produced by a script show denois tuning. The x stands for the ground truth image and x denotes the image ˆ restored from noisy one x using (7. Figure 7.6 shows examples of ground truth images. m∗ ) was selected out of a prescribed set Θ = {(σ1 . . images restored by the linear PCA and kernel PCA. The best parameters (σ ∗ . . noisy images. The greedy kernel PCA Algorithm 3. An example of denoising of the testing data of USPS is implemented in script show denoising. Having the parameters tuned the resulting kernel PCA and linear PCA are trained on the whole training set. The function kpcarec is used to denoise images using the kernel PCA model. εM S (σ. 89 .4 is used for training the model. The training for linear PCA is implemented in the script train lin denois.
15 0.1 0.25 0.6 0.5 0. right_button − erase.45 0. middle_button − classify.Control: left_button − draw.2 0.35 0.05 0 0 0.4 0.5 0.8 0.1: Figure (a) shows handwritten training examples of numeral ’1’ and Figure (b) shows corresponding normalized images of size 16 × 16 pixels. 90 . 0.7 0.9 1 (a) (b) Figure 7.3 0.2 0.3 0.1 0.4 0.
3 0.1 0.2 0.4 0. middle_button − classify.4 0.5 0.45 0.3 0.35 0. 0.5 0.8 0.3 0.3 0.05 0 0 1 2 3 4 5 6 7 8 9 0 4 5 6 0. Figure (a) shows handwritten numerals and Figure (b) shows recognized numerals.2 0.2 0.5 0.05 0 0 0.6 0.6 0.35 0.45 0.25 0.25 0.1 0.2 0.9 1 (b) Figure 7.7 0.15 0.15 0.9 1 (a) 0.Control: left_button − draw. 91 .1 0.4 0.5 0. right_button − erase.4 0 3 2 0.1 0.8 1 9 8 0.7 0.2: A snapshot of running OCR system.
3: Demonstration of idea of image denoising by kernel PCA. Distortion model KPCA model θ∈Θ Figure 7.7 6 5 4 3 2 1 0 −1 −2 −3 −4 −4 Ground truth Noisy examples Corrupted Reconstructed −3 −2 −1 0 1 2 3 4 5 6 7 Figure 7.4: Schema of tuning parameters of kernel PCA model used for image restoration. Preimage problem error θ Data generator Error assessment 92 .
18 20 dim = 50 dim = 100 dim = 200 dim = 300 17 18 16 16 15 14 εMS 14 εMS 12 10 8 6 3.5 13 12 11 10 0 10 20 30 40 50 dim 60 70 80 90 100 4 4.5 σ 6 6.5: Figure (a) shows plot of εM S with respect to output dimension of linear PCA. 93 . Figure(b) shows plot of εM S with respect to output dimension and kernel argument of the kernel PCA. Ground truth Numerals with added Gaussian noise Linear PCA reconstruction Kernel PCA reconstruction Figure 7.5 5 5.6: The results of denosing the handwritten images of USPS database.5 8 (a) (b) Figure 7.5 7 7.
• AdaBoost and related learning methods. The future planes involve the extensions of the toolbox by the following methods: • Feature selection methods. • Discrete probability models estimated by the EM algorithm. We welcome anyone who likes to contribute to the STPRtool in any way. 94 . • Eﬃcient optimizers to for large scale Support Vector Machines learning problems. suggestion or other feedback which would help in the development of the STPRtool. There are many relevant methods and proposals which have not been implemented yet.Chapter 8 Conclusions and future planes The STPRtool as well as its documentation is being updating. • Probabilistic Principal Component Analysis (PPCA) and mixture of PPCA estimated by the ExpectationMaximization (EM) algorithm. Therefore we are grateful for any comment. Please send us an email and your potential contribution. • Model selection methods for SVM.
3th edition. Generalized discriminant analysis using a kernel approach. Duin. Duda. Journal of the Royal Statistical Society. Multiclass support vector machine. 2003. a [9] V. IEEE Computer Society.E. Baudat and F. [4] N. Oxford.M. edition. 2nd. 2001.B. Laurendeau. Cesk´ vysok´ uˇen´ technick´. N. Cristianini and J. Prtools: A matlab toolbox for pattern recognition..M. Anals of Mathematical Statistics. and Suen C. [3] C. Hlav´ˇ.Bibliography [1] T. 36(9):1985–1996. In R. Maximum likelihood from incomplete data via the EM Algorithm. 1977. 2002. 2000. Pattern Recognition. 16th International Conference on Pattern Recognition. ac [10] V. and D. June 1962. Bishop. Cambridge University Press. Katedra kybernetiky. An iterative algorithm learning the maximal margin classiﬁer. Hart. 2000.R. [8] V. and D. 12(10):2385–2404. Laird. Kasturi. 95 . [6] R. 39:185– 197. [7] R. In Czech). Master’s thesis. Clarendon Press.W. 33:420–431. Neural Networks for Pattern Recognition. John Wiley & Sons. pages 236–239. Rubin. volume 2. Franc and V. Stork. 1997. February 2000.P.G. [5] A. Programov´ n´stroje pro rozpozn´v´n´ (Pattern Recognition Programe a a a ı ˇ e ming Tools. Support Vector Machines. Fakulta e c ı e elektrotechnick´. Dempster. ShaweTaylor. [2] G. Neural Computation. Hlav´ˇ. Franc and V.W. P. Franc.O. Pattern Classiﬁcation. ac D. Great Britain. Anderson and R.P. 2000. editors. Bahadur. Anouar. Classiﬁcation into two multivariate normal distributions with diﬀerrentia covariance matrices.
Henderson. MIT Press. The EM Algorithm and Extensions. New York. Bhattacharya. editors. M¨ ller.A. Springer.J Jackel. Solla. S.K.J. Computer Analysis of Images and Patterns. Hu. 96 .. March 2002.R. Kernel hebbian o algorithm for singlefram superresolution. and D. MA. Tsang. 2000. and L. In Y. The preimage problem in kernel methods.K.A. B. S. G. M¨ ller. Probabilities for sv machines.W. Germany. W. New York. A comparison of methods for multiclass support vector machins. Platt. Springer. 1989. Weston. and K. K. B. Sch¨lkopf. January 2000. A. Fisher discriminant analysis with kernel.H. D. Hubbard. USA. May 2004.S. Kwok and I. Douglas. Springer.C. B. Denker. [14] S.S.J. ECCV Workshop. [21] I. In A. Nabney. Kernel o u a pca and denoising in feature spaces. and K.. Howard.R. In Proceedings of the Twentieth International Conference on Machine Learning (ICML2003). Mika. Keerthi. Cambridge. R¨tsch.A. and D. IEEE Transactions on Neural Networks. editors. IEEE Transactions on Neural Networks. Sch¨lkpf. Greedy algorithm for a training set reduction in the ac kernel methods. and S. IEEE. Advances in pattern recognition. In. In Leonardis Aleˇ and Bischof Horst.T. [20] S. Statisical Learning in Computer Vision. a o u [19] S. Franz Matthias. Cohn. John Wiley & Sons. 11(1):124–136. O. pages 41–48. [22] J.[11] V. In N. Scholz. and Sch¨lkopf Bernhard. Mika. E. Neural Computation. Principal Component Analysis. Advances in Large Margin Classiﬁers (Neural Information Processing Series). J. Neural Networks for Signal Processing.T. Washington. pages 408–415. pages 536 – 542.W. 2003.J. Krishnan. A fast iterative nearest point algorithm for support vector machine classiﬁer design. Backpropagation applied to handwritten zip code recognition. J.S. Wilson. NETLAB : algorithms for pattern recognition. [15] Kim Kwang. editors. Larsen. M. Murthy. Petkov and M. Smola. 1986. In M.E. [18] G. J. [12] C. 13(2). s editors. Hsu and C. MIT Press.T. Lin. 1997. Jollife. Shevade. [17] Y. P. Hlav´ˇ. Schuurmans. C. and G. August 2003. D. McLachlan and T. Boser. 1999. LeCun. editors. pages 426–433. Westenberg. R¨tsch. 2002. Berlin. Kearns. [13] I. Bartlett. B. Scholkopf. SpringerVerlag. R. [16] J. Advances in Neural Information Processing Systems 11. Franc and V. 1:541–551. 1999. Smola. London.
A. 56:409–456.J. Nonlinear component analysis as a kernel eigenvalue problem. Royal Statistical Soc. editors. Advances in Kernel Methods – Support o Vector Learning. Ahler. Learning with Kernels.I. 2:81–88. Knirsch. Microsoft Research. Mustererkennung 199820. 1998. o [31] V. The MIT Press. Neural Computation. 1998.R. P.J. Fast approximation of support o vector kernel expansions. Burges. Germany. Kibernetika. Technical Report MSRTR9814. Smola. Sch¨lkopf and A. [30] B. Sequential minimal optimizer: A fast algorithm for training support vector machines. Burges.C. Riply. Neural networks and related methods for classiﬁcation (with discusion). DAGM. [27] B. Platt. pages 124–132. 1968. Smola. and F. and C. M.[23] J. [25] M. 1995. Ten lectures on statistical and structural pattern ac recognition. Levi. 2002. Berlin. MA. MIT Press. Schanz. Cambridge. Smola. 1999. [24] B. Schlesinger. A connection between learning and selflearning in the pattern recognition (in Russian). Muller.. Hlav´ˇ. J. R. In P. 2002. [28] B. The nature of statistical learning theory. Schlesinger and V. Springer Verlag. 1998. Redmond. May.I. Series B. 10:1299–1319. and K. Sch¨lkopf.D. C. SpringerVerlag. Kluwer Academic Publishers. Scholkopf. and A. and an interpretation of clustering as approximation in feature spaces. 1994. 97 . [29] B. A.J. Sch¨lkopf. Smola. Vapnik. [26] M.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.