-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmy_first_Net.py
116 lines (94 loc) · 4.52 KB
/
my_first_Net.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
import numpy as np
import matplotlib.pyplot as plt
def Sigmoid(x): # シグモイド関数
return 1 / (1 + np.exp(-x))
def dif_Sigmoid(x): # シグモイド関数の微分
fx = Sigmoid(x)
return fx * (1 - fx)
class my_Neural_Network:
def __init__(self): # 重みとバイアスの初期化
self.w1,self.w2,self.w3,self.w4,self.w5,self.w6= 1,1,1,1,1,1
self.b1,self.b2,self.b3 = 0,0,0
def feedforward(self,x): # 順伝播 モデルの構造
h1 = Sigmoid(x[0] * self.w1 + x[1] * self.w2 + self.b1)
h2 = Sigmoid(x[0] * self.w3 + x[1] * self.w4 + self.b2)
out = Sigmoid(h1 * self.w5 + h2 * self.w6 + self.b3)
return out
def train(self,train_data,ture_y): # 逆伝播
learn_rate = 0.1
epochs = 2001
x_epoch = []
y_loss = []
for epoch in range(epochs):
for x,y_true in zip(train_data,ture_y):
sum_h1 = self.w1 * x[0] + self.w2 * x[1] + self.b1
h1 = Sigmoid(sum_h1)
sum_h2 = self.w3 * x[0] + self.w4 * x[1] + self.b2
h2 = Sigmoid(sum_h2)
sum_out = self.w5 * h1 + self.w6 * h2 + self.b3
out = Sigmoid(sum_out)
y_pred = out
d_L_d_y_pred = -2 * (y_true - y_pred)
d_y_pred_d_h1 = self.w5 * dif_Sigmoid(sum_out)
d_y_pred_d_h2 = self.w6 * dif_Sigmoid(sum_out)
d_y_pred_d_w5 = h1 * dif_Sigmoid(sum_out)
d_y_pred_d_w6 = h2 * dif_Sigmoid(sum_out)
d_y_pred_d_b3 = dif_Sigmoid(sum_out)
d_h1_d_w1 = x[0] * dif_Sigmoid(sum_h1)
d_h1_d_w2 = x[1] * dif_Sigmoid(sum_h1)
d_h1_d_b1 = dif_Sigmoid(sum_h1)
d_h2_d_w3 = x[0] * dif_Sigmoid(sum_h2)
d_h2_d_w4 = x[1] * dif_Sigmoid(sum_h2)
d_h2_d_b2 = dif_Sigmoid(sum_h2)
self.w1 -= learn_rate * d_L_d_y_pred * d_y_pred_d_h1 * d_h1_d_w1
self.w2 -= learn_rate * d_L_d_y_pred * d_y_pred_d_h1 * d_h1_d_w2
self.w3 -= learn_rate * d_L_d_y_pred * d_y_pred_d_h2 * d_h2_d_w3
self.w4 -= learn_rate * d_L_d_y_pred * d_y_pred_d_h2 * d_h2_d_w4
self.w5 -= learn_rate * d_L_d_y_pred * d_y_pred_d_w5
self.w6 -= learn_rate * d_L_d_y_pred * d_y_pred_d_w6
self.b1 -= learn_rate * d_L_d_y_pred * d_y_pred_d_h1 * d_h1_d_b1
self.b2 -= learn_rate * d_L_d_y_pred * d_y_pred_d_h2 * d_h2_d_b2
self.b3 -= learn_rate * d_L_d_y_pred * d_y_pred_d_b3
if epoch % 100 == 0: # 100回訓練ごとLossをプリントする
y_preds = np.apply_along_axis(self.feedforward, 1, train_data)
loss = ((true_y - y_preds) ** 2).mean()
print("Epoch %d loss: %.3f" % (epoch, loss))
x_epoch.append(epoch)
y_loss.append(loss)
if epoch % 500 == 0: # 500回ごと重みとバイアスをプリントする
print("weight w1 %.3f w2 %.3f w3 %.3f w4 %.3f w5 %.3f w6 %.3f" % (
self.w1, self.w2, self.w3, self.w4, self.w5, self.w6))
print("bias b1 %.3f b2 %.3f b3 %.3f" % (self.b1, self.b2, self.b3))
plt.plot(x_epoch, y_loss)
plt.xlabel("epoch")
plt.ylabel("loss")
def test(train,x): # 訓練したのニューラルネットワークに身長と体重データを入れて男性と女性を判断する
h1 = Sigmoid(x[0] * train.w1 + x[1] * train.w2 + train.b1)
h2 = Sigmoid(x[0] * train.w3 + x[1] * train.w4 + train.b2)
out = Sigmoid(h1 * train.w5 + h2 * train.w6 + train.b3)
print("予測結果（１に近くは男性、０に近くは女性）：",out)
train_data = np.array([[5, 11], # [身長、体重]と平均の偏差、身長体重偏差高いの方が男性、低いの方が女性
[12, 8],
[22,11],
[6, 2],
[25, 6],
[-7, 1],
[-15, -6],
[1, 2],
[-9,0],
[-13,-2]])
true_y = np.array([1, # １は男性、０は女性
1,
1,
1,
1,
0,
0,
0,
0,
0,])
test = [11,2]
network = my_Neural_Network()
network.train(train_data, true_y)
network.test(test)
plt.show()