You are on page 1of 14

ANFIS, Adaptive Neuro-Fuzzy Inference System,

- . ANFIS

Tagaki-Sugeno.

(Nedjah, Nadia), .
IF-THEN (-)
- (Nedjah, Nadia pp. 5383).
ANFIS
, .

,
,
. ,

ANFIS


, Henri Poincar. Poincar,
:
.
.
Poincar .

.

.

.
; Poincar
.
:

.
.

.
, .
(
).
:
(context).
( 1) ,
Poincar -

(-) . (Mitchell 2009 pp.19-22)

1 : Henri Poincar



. ;
.
, (
), .

,
.

().
.
Poincar
, Poincar
.


. ( 2)

2 :



.

. ,

.

.
.
.
,
. 41,
:

, ,
. (Stewart & Cohen 1997 pp.49-52)


.

,

.


.
.
(Low Level
Approach). :
,
( ) ..
,
.

. (mediamax.gr, 2011)
( 1)
-
.

, .
, .

, , ,
( 4). ( 2001 20)
4: '


.
( 2001 26):
1:

:
..
.
(

)
..
. ( )
, .
. ..
(.. ),
( ),
.
(universal
approximators) (data
fitting).

. .
,
. ,
.
.
.


(fuzzy logic) .
,
. , .
, 2500 ,
0-1, -.
:
, .. 300.
270.
.
, , .. 20%.
,
, ,
.
.
4.
, .
,
. :


, .

, .. 5: , , , ,
.
:
; 62%
10% .
!

.
. .
, ,
.
:

,

( )
.

.
.
(Expert systems).
Sugeno
ANFIS
. ANFIS
Sugeno e Tsukamoto
.
x y
z. Sugeno :
1. x A1 y B1, f1=p1x+q1y+r1
2. x A2 y B2, f2=p2x+q2y+r2
:

1 l,i i . i
:
Ol,i=Ai(x) i =1,2
Ol,i=Bi2(x) i =3,4
To x( y) i Ai ( Bi2)
( ) Ol,i
(A1,A2,B1,B2)
:
ai, bi, ci
.

2 Prod.
.

.
AND .

4 Norm . O
i (firing strength)
i .

(normalized firing
strengths)
4 i :

wi 3 {pi. qi, ri}



(consequent parameters)
5 sum,
:

Sugeno, .

:


ANFIS
Jang, Sun Mizutani.
9.4.2 Computational intelligence paradigms : theory &
applications using MATLAB S. Sumathi Surekha Paneerselvam.
(forward pass)
(backward pass).

4.
.
2: ANFIS






ANFIS
,
-

Stone-Weierstrass:
D F
-, D :
: f(x)=1 F
: x1x2 D, f F,

f(x1)f(x2).
: f g F,
fg af+bg F a b.
F C(D),
- D.
>0 g C(D), f F
g(x)-f(x) < xD
ANFIS .
ANFIS MATLAB
ANFIS MATLAB
anfisedit.
m-file. MATLAB
, , ANFIS
ANFIS MATLAB ( 2009):

ANFIS. Generate FIS.


FIS=GENFIS1(DATA)
Sugeno (Fuzzy Inference System - FIS)
( ). FIS
ANFIS. DATA
+1 FIS,
.

GENFIS1
gbell/mf .
.
FIS=GENFIS1(DATA, NUMMFS, INPUTMF,
OUTPUTMF). :
fis = genfis1(data, [3 7], char(pimf, trimf));
data = [rand(10,1) 10*rand(10,1)-5 rand(10,1)];
fis = genfis1(data, [3 7], char(pimf,trimf));
[x,mf] = plotmf(fis,input,1);
subplot(2,1,1), plot(x,mf);
xlabel(input 1 (pimf));
[x,mf] = plotmf(fis,input,2);
subplot(2,1,2), plot(x,mf);
xlabel(input 2 (trimf));
:

GENFIS2 FIS Sugeno



.



.
ANFIS

Sugeno, .
:
[FIS,ERROR] = ANFIS(TRNDATA)
[FIS,ERROR] = ANFIS(TRNDATA,INITFIS)
EVALFIS
. Y = EVALFIS (U, FIS)
U
Y.
ANFIS.

:
, 2001

,
Nedjah, Nadia, ed.
Studies in Fuzziness and Soft Computing.
Germany: Springer Verlag.
Jang, Sun, Mizutani (1997)
Neuro-Fuzzy and Soft Computing
Prentice Hall, pp 335-368
Stewart Ian, Cohen Jack 1997
Figments of Reality
Cambridge: Cambridge University Press
Sumathi S. & Sureha P. 2010
Computational Intelligence Paradigms: Theory and Applications using MATLAB
New York: Taylor & Francis Group
Mitchell, Melanie 2009
Complexity, A Guided Tour
Oxford: Oxford University Press

http://www.mediamax.gr/articles/inDetail.cfm?EntryID=31 (3/12/2011)

You might also like