-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtwolayernet.py
192 lines (157 loc) · 6.85 KB
/
twolayernet.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
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
from needs import relu, sigmoid, softmax, cross_entrppy_error, cal_gradient, sigmoid_grad
from needs import Relu, Affine, Softmax_With_Loss, Sigmoid
from collections import OrderedDict
import numpy as np
class TwoLayerNet_Pro: # 真正意义上的网络
def __init__(self, input_size, hidden_size, output_size, w=0.01):
self.params = {} # 随机生成初始值，W1和W1权重值使用高斯随机分布生成，偏置b1和b2全为0
self.params['W1'] = w * np.random.randn(input_size, hidden_size)
self.params['b1'] = np.zeros(hidden_size)
self.params['W2'] = w * np.random.randn(hidden_size, output_size)
self.params['b2'] = np.zeros(output_size)
# 创建 生成层
self.layers = OrderedDict()
# 第一仿射层，a = W * x + b
self.layers['Affine1'] = Affine(self.params['W1'], self.params['b1'])
# # 第一激活层，Relu层
self.layers['Relu1'] = Relu()
# self.layers['Sigmoid1'] = Sigmoid()
# 第二仿射层
self.layers['Affine2'] = Affine(self.params['W2'], self.params['b2'])
# 神经网络最后一层
self.lastLayer = Softmax_With_Loss()
def loss(self, x, t):
y = self.predict(x)
return self.lastLayer.forward(y, t)
def predict(self, x):
for layer in self.layers.values(): # 分别计算Afffine1、Relu1、Affine2层
x = layer.forward(x)
return x # x = a2(z2 = softmax(a2))
def cal_accuracy(self, x, t):
y = self.predict(x)
y = np.argmax(y, axis=1)
if t.ndim != 1:
t = np.argmax(t, axis=1) # t若不是1维，进行正规化
return np.sum(y == t) / float(len(t))
def gradient(self, x, t):
# forward
self.loss(x, t)
# backward
dout = 1
dout = self.lastLayer.backward(dout)
layers = list(self.layers.values())
layers.reverse() # 列表反向
for layer in layers: # 反向计算各层的导数
dout = layer.backward(dout)
grads = {}
grads['W1'] = self.layers['Affine1'].dW
grads['b1'] = self.layers['Affine1'].db
grads['W2'] = self.layers['Affine2'].dW
grads['b2'] = self.layers['Affine2'].db
return grads
class TwoLayerNet():
def __init__(self, input_size, hidden_size, output_size, w=0.01):
self.params = {} # 随机生成初始值，W1和W1权重值使用高斯随机分布生成，偏置b1和b2全为0
self.params['W1'] = w * np.random.randn(input_size, hidden_size)
self.params['b1'] = np.zeros(hidden_size)
self.params['W2'] = w * np.random.randn(hidden_size, output_size)
self.params['b2'] = np.zeros(output_size)
def predict(self, x): # 检测函数，返回检测结果
W1, W2 = self.params['W1'], self.params['W2']
b1, b2 = self.params['b1'], self.params['b2']
a1 = np.dot(x, W1) + b1
# z1 = sigmoid(a1)
z1 = relu(a1)
a2 = np.dot(z1, W2) + b2
y = softmax(a2)
return y
def loss(self, x, t): # 计算损失函数的值
y = self.predict(x)
return cross_entrppy_error(y, t)
def numerical_differential(self, x, t): # 数值微分方法计算梯度
def loss_W(W): return self.loss(x, t)
grads = {}
grads['W1'] = cal_gradient(loss_W, self.params['W1'])
grads['b1'] = cal_gradient(loss_W, self.params['b1'])
grads['W2'] = cal_gradient(loss_W, self.params['W2'])
grads['b2'] = cal_gradient(loss_W, self.params['b2'])
return grads
def gradient(self, x, t): # 误差反向传播算法高速计算梯度
W1, W2 = self.params['W1'], self.params['W2']
b1, b2 = self.params['b1'], self.params['b2']
grads = {}
batch_num = x.shape[0]
# forward
a1 = np.dot(x, W1) + b1
z1 = sigmoid(a1)
a2 = np.dot(z1, W2) + b2
y = softmax(a2)
# backward
dy = (y - t) / batch_num
grads['W2'] = np.dot(z1.T, dy)
grads['b2'] = np.sum(dy, axis=0)
da1 = np.dot(dy, W2.T)
dz1 = sigmoid_grad(a1) * da1
grads['W1'] = np.dot(x.T, dz1)
grads['b1'] = np.sum(dz1, axis=0)
return grads
def cal_accuracy(self, x, t):
y = self.predict(x)
y = np.argmax(y, axis=1)
t = np.argmax(t, axis=1)
return np.sum(y == t) / float(len(t))
class TwoLayerNet_Pro2:
def __init__(self, input_size, hidden_size1, hidden_size2,hidden_size3, output_size, w=0.01):
self.params = {} # 随机生成初始值，W1和W1权重值使用高斯随机分布生成，偏置b1和b2全为0
self.params['W1'] = w * np.random.randn(input_size, hidden_size1)
self.params['b1'] = np.zeros(hidden_size1)
self.params['W2'] = w * np.random.randn(hidden_size1, hidden_size2)
self.params['b2'] = np.zeros(hidden_size2)
self.params['W3'] = w * np.random.randn(hidden_size2, hidden_size3)
self.params['b3'] = np.zeros(hidden_size3)
self.params['W4'] = w * np.random.randn(hidden_size3, output_size)
self.params['b4'] = np.zeros(output_size)
# 创建 生成层
self.layers = OrderedDict()
# 第一仿射层，a = W * x + b
self.layers['Affine1'] = Affine(self.params['W1'], self.params['b1'])
# # 第一激活层，Relu层
self.layers['Relu1'] = Relu()
# self.layers['Sigmoid1'] = Sigmoid()
# 第二仿射层
self.layers['Affine2'] = Affine(self.params['W2'], self.params['b2'])
self.layers['Relu2'] = Relu()
self.layers['Affine3'] = Affine(self.params['W3'], self.params['b3'])
self.layers['Relu3'] = Relu()
self.layers['Affine4'] = Affine(self.params['W4'], self.params['b4'])
# 神经网络最后一层
self.lastLayer = Softmax_With_Loss()
def loss(self, x, t):
y = self.predict(x)
return self.lastLayer.forward(y, t)
def predict(self, x):
for layer in self.layers.values(): # 分别计算
x = layer.forward(x)
return x # x = a2(z2 = softmax(a2))
def cal_accuracy(self, x, t):
y = self.predict(x)
y = np.argmax(y, axis=1)
if t.ndim != 1:
t = np.argmax(t, axis=1) # t若不是1维，进行正规化
return np.sum(y == t) / float(len(t))
def gradient(self, x, t):
# forward
self.loss(x, t)
# backward
dout = 1
dout = self.lastLayer.backward(dout)
layers = list(self.layers.values())
layers.reverse() # 列表反向
for layer in layers: # 反向计算各层的导数
dout = layer.backward(dout)
grads = {}
grads['W1'] = self.layers['Affine1'].dW
grads['b1'] = self.layers['Affine1'].db
grads['W2'] = self.layers['Affine2'].dW
grads['b2'] = self.layers['Affine2'].db
return grads