06.03.2013 Views

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

x 1<br />

x 2<br />

x 1<br />

x 2<br />

The training process of an ADALINE neuron is realized by the Least<br />

Means Square (LMS) algorithm presented below.<br />

Procedure LMS<br />

Input d i <strong>and</strong> X k<br />

Output weight vector W k ~<br />

Begin<br />

R<strong>and</strong>omize initial weight vector W0 ;<br />

Repeat<br />

T<br />

∈ = d − X w ;<br />

k k k k<br />

w =<br />

/ | |<br />

1 w + α<br />

k + k ∈k<br />

X k X k<br />

~<br />

Until<br />

Print<br />

End.<br />

∑w i x i<br />

∑wi xi<br />

~<br />

mod<br />

w ;<br />

k<br />

~<br />

~<br />

First layer<br />

~<br />

~<br />

| ( w − ) | pre defined<br />

k 1 w < −<br />

+ k<br />

~<br />

~<br />

~<br />

~<br />

2<br />

;<br />

y1<br />

y 2<br />

s1<br />

s 2<br />

limit;<br />

∑si yi<br />

Second layer<br />

Fig. 14.11: A network of three ADALINES in two layers for realizing an ‘XOR’ gate.<br />

Z

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!