Biraz geç oldu ama kusruma bakma yazmaya çalışayım:
1)for an ANN with one neuron, process four step for Hebb learning rule
W1=[-1 1] c=1
x1=[1 -2] x2[0 1] x3=[2 3] x4[1 -1]
a) f(u)=1/(1+e^(-u))
b f(u)=sgn(u)
2)Use Perceptron learning rule for training network
W1=[0 1 0] c=1
f(u)=sgn(u)
x1=[-1 1 2] d1=-1
x2=[-1 -1 0] d2=1
3)an ANN with one neuron is trained for the following sets with perceptron learning Rule
X1=[1 3 -2 -1] d1=-1
X2=[0 2 -1 -1] d2=1
X3=[-2 -3 0 -1] d3=-1
And adapted wiegth vector is obtained as below
W4=[3 2 6 1]
for c=1 and f(u)=sgn(u) find the W3,W2,W1 by back-tracking the training
Istanbul Universitesi
Bilgisayar Mühendisligi
|
Del.icio.us
Digg
Facebook
Furl
Google
Blink
Simpy
Spurl
Y! MyWeb