import?numpy?as?np
import?math?
‘‘‘
輸入層矩陣
layer1out???a??x?1
??????X1??X2??X3??X4???X5?...?Xa?????T
1

隱藏層矩陣
layer2????b?x?a
?????w1?w2?w3?w4?w5?......wa
1
2
...
b

layer2tempout????b?x?1?????
??????y1?y2?y3?y4?y5?.....?yb???????????T
1

layer2out??b?x?1
???????y1?y2?y3?y4?y5?.....?yb???????????T
1


輸出層矩陣
layer3????c?x??b?
???????v1?v2?v3?v4?v5?......vb
1
2
...
c

layer3tempout???c???x?1
??????z1?z2?z3?z4?z5?....?zc????????????T
1

layer3out????c?x?1
?????z1??z2??z3??z4?z5?....zc
1

目標值
target????c??x?1
?????t1?t2?t3?t4?t5?tc?????????T
1
‘‘‘
def?sigmoid（x）:
??????return?1?/?（1?+?np.exp（-x））
????
def?sigmoid_der（x）:
????return?sigmoid（x）*?（1-?sigmoid（x））

#輸入層，隱藏層，輸出層，目標值?學習速度，迭代次數，收斂值
def?Back_Propagation（ly1out?ly2?ly3?targetalpha