You are on page 1of 25

End to End Machine Learning Project : From Data to

Insights

1
2
Session 5 : Content

✔ Orthogonal Projection
✔ Transpose of Vector & Matrix : Row-Column
✔ Reverse Matrix
✔ Sets , subsets , power sets
✔ Union, Intersection , Compliment & Difference
✔ Countable : Indexed ; Uncountable : Infinite elements uncountable
✔ Discrete set vs Continuous set : Set of parameters in DL is discrete
✔ Sequences
✔ Analysis vs Calculus
✔ Variable vs Multivariate Calculus vs
✔ Derivatives as rate of change

3
Chapter 2.1

Orthogonal Projections

4
Chapter 2.1

Orthogonal Projection
● Orthogonal Projection
● A generalization of the notion of perpendiculaity
● The orthogonal projection minimizes the distance.
● Linear regression is an orthogonal projection.

5
6
Chapter 2.1

Inverse of a matrix
● Inverse of a matrix is just like inverse of numbers .
● inverse of 3 is ⅓
● Inverse of a matrix exist if it’s square matrix

7
8
Chapter 2.1

Transpose of Vector & Matrix


● Always exists
● Row vector is a transpose of column vector

9
10
Chapter 2.1

Orthogonal Matrix
● An orthogonal matrix is a matrix for which inverse = transpose
● If our matrix is orthogonal, we can easily compute its inverse
which is the transpose without Going through formula which is
computationally expensive for large matrix.

11
Chapter 2.2

Analytic & Discrete Mathematics in Artificial Intelligence


● Sets , sequences
● Functions , Derivatives

12
Chapter 2.1

Sets , Subsets & Power sets


● Sets are discrete collection of objects.
● Subsets : If all the elements of Set A are also present in Set B.
● Power Set : the set of all subsets of the Set A including the Set
itself and the null or empty set.

13
Chapter 2.1

Sets , Subsets & Power sets


● We will this examples in probabability
● In statistics Discrete Variables Take values in Sets

14
Power set

15
16
Chapter 2.1

Countable vs Uncountable set


● Countable : Indexed ; Uncountable : Infinite elements
uncountable
● Countable set is a set whose elements we can count
● A set is countable bcz we can index its elements
● A list is countable bcz we can index its elements.

17
Chapter 2.1

Notion of Discrete
● Discrete is countable
● Characterized by number of generally
● Set of parameters in ML DL is discrete and unique .

18
19
Chapter 2.1

Notion of Sequence
● Sets of numbers that follow a pattern or a rule
● Rule could be a function , A mathematical Operation
application
● Mathematical Operations we have Arithmetic & Geometric
● Arithmetic is about + & -, Geometric is abou * & /
● In AI we have what we call sequence DATA
20
Chapter 2.1

Notion of Sequence
Sequence data is data that has a certain sequence , enchainment
by a rule
● String ="Samsung innovation camp" Is a sequence of
characters
● Stock prices are a sequence of data points in time ,price in
time.

21
Chapter 2.1

Notion of Sequence
● Text
● Aauudio
● Signals
● Time series data

22
23
Chapter 2.1

Analysis Vs Calculus
● Analysis is about studying the analytical things we can do
inside constructions that algebra has created.
● Analytical Means studying their properties
● Calculus is about calculations of these notions like calcul
of limits & derivatives .

24
Chapter 2.1

Variable vs Multivariate Calculus


● Input of a one var function is one param unlike multivar .
● We define limits for 1 var of multivar functions
● we define continuity
● we define derivatives and integration

25

You might also like