책없을때 보려고 블로그에 백업 >_<
Step 1 : Intialize weights and counter
Step 2 : Set learning rate a(0 <= a <= 1)
Step 3 : For each training pattern pair (x, d)
Step 5 : Compare output and desired output
Step 6 : Update weights
Step 7 : Increase counter and goto Step 3
Step 8 : Test stop condition
w <- 0 or small random value
p <- number of training pattern paris
k <- 1
p <- number of training pattern paris
k <- 1
Step 2 : Set learning rate a(0 <= a <= 1)
Step 3 : For each training pattern pair (x, d)
do Step 4 - 7 until k = p
Step 4 : Compare output
Step 4 : Compare output
y = f(NET)
+1 ; NET > T
= 0 ; NET = T
-1 ; NET < T
+1 ; NET > T
= 0 ; NET = T
-1 ; NET < T
Step 5 : Compare output and desired output
If y = d, k <- k + 1 and goto Step 3
Step 6 : Update weights
delta_w <- a * (dk - yk) * xk
w <- w + delta_w
w <- w + delta_w
Step 7 : Increase counter and goto Step 3
Step 8 : Test stop condition
If no weights changed in Step 3 - 7, stop
else, k <- 1 and goto Step 3
'신경망' 카테고리의 다른 글
| 모멘텀 BP 알고리즘 momentum BP algorithm (0) | 2009/01/15 |
|---|---|
| Backpropagation algorithm (0) | 2008/12/30 |
| MADALINE 학습 알고리즘 - MR I (4) | 2008/11/28 |
| 퍼셉트론 학습 알고리즘 (0) | 2008/11/28 |
| 퍼셉트론 Perceptron 패턴인식 데모 (8) | 2008/11/04 |
| Perceptron 시뮬레이션 (2) | 2008/10/21 |
댓글을 달아 주세요