关于python中的takenoarguments的解决⽅法针对第四章编写的代码出现的错误做⼀个总结
Traceback (most recent call last):
File "H:\image\chapter4\p81_chongxie.py", line 160, in <module>
l1 = Linear(X, W1, b1)
TypeError: Linear() takes no arguments
出问题时的init⽅法的图⽚
可以看出init两边只有⼀个下划线 _.
解决办法：把init的两边改成两个下划线 __。即可。
代码运⾏环境：win7系统 + anaconda3_2020
第四章的代码如下：
1#######################数据结构部分#############################################
2import numpy as np
3import matplotlib.pyplot as plt
4
5# %matplotlib inline
6
7class Node(object):
8def__init__(lf, inbound_nodes = []):
9 lf.inbound_nodes = inbound_nodes
10 lf.value = None
11 lf.outbound_nodes = []
12
13 lf.gradients = {}
14
15for node in inbound_nodes:
16 node.outbound_nodes.append(lf)
17
18def forward(lf):
19rai NotImplementedError
20
21def backward(lf):
22rai NotImplementedError
23
24
25class Input(Node):
26def__init__(lf):
27 Node.__init__(lf)
28
29def forward(lf):
30pass
31
32def backward(lf):
33 lf.gradients = {lf : 0}
34for n in lf.outbound_nodes:
35 lf.gradients[lf] += n.gradients[lf]
36
37##################################################################################
38class Linear(Node):
39def__init__(lf, X, W, b):
40 Node.__init__(lf, [X, W, b])
41
42def forward(lf):
43 X = lf.inbound_nodes[0].value
44 W = lf.inbound_nodes[1].value
45 b = lf.inbound_nodes[2].value
46 lf.value = np.dot(X, W) + b
47
48def backward(lf):
49 lf.gradients = {n: np.zeros_like(n.value) for n in lf.inbound_nodes }
50for n in lf.outbound_nodes:
51 grad_cost = n.gradients[lf]
52 lf.gradients[lf.inbound_nodes[0]] += np.dot(grad_cost, lf.inbound_nodes[1].value.T)
53 lf.gradients[lf.inbound_nodes[1]] += np.dot(lf.inbound_nodes[0].value.T, grad_cost)
54 lf.gradients[lf.inbound_nodes[2]] += np.sum(grad_cost, axis = 0, keepdims = Fal)
55
56################################################################################### 57class Sigmoid(Node):
58def__init__(lf, node):
59 Node.__init__(lf, [node])
60
61def _sigmoid(lf, x):
62return 1. / (1. + np.exp(-x)) #exp() ⽅法返回x的指数,e的x次幂
63
64def forward(lf):
65 input_value = lf.inbound_nodes[0].value
66 lf.value = lf._sigmoid(input_value)
67
68def backward(lf):
69 lf.gradients = {n: np.zeros_like(n.value) for n in lf.inbound_nodes}
70for n in lf.outbound_nodes:
71 grad_cost = n.gradients[lf]
72 sigmoid = lf.value
73 lf.gradients[lf.inbound_nodes[0]] += sigmoid * (1 - sigmoid) * grad_cost
74
75
76class MSE(Node):
77def__init__(lf, y, a):
78 Node.__init__(lf, [y, a])
79
80
81def forward(lf):
82 y = lf.inbound_nodes[0].shape(-1, 1)
83 a = lf.inbound_nodes[1].shape(-1, 1)
84
85 lf.m = lf.inbound_nodes[0].value.shape[0]
86 lf.diff = y - a
87 lf.value = np.mean(lf.diff**2)
88
89
90def backward(lf):
91 lf.gradients[lf.inbound_nodes[0]] = (2 / lf.m) * lf.diff
92 lf.gradients[lf.inbound_nodes[1]] = (-2 / lf.m) * lf.diff
93
94
95
96##########################计算图部分#############################################
97def topological_sort(feed_dict):
98 input_nodes = [n for n in feed_dict.keys()]
99 G = {}
100 nodes = [n for n in input_nodes]
101while len(nodes) > 0:
102 n = nodes.pop(0)
103if n not in G:
104 G[n] = {'in' : t(), 'out' : t()}
105for m in n.outbound_nodes:
106if m not in G:
107 G[m] = {'in' : t(), 'out' : t()}
108 G[n]['out'].add(m)
109 G[m]['in'].add(n)
110 nodes.append(m)
111
112 L = []
113 S = t(input_nodes)
114while len(S) > 0 :
115 n = S.pop()
116if isinstance(n, Input):
117 n.value = feed_dict[n]
118 L.append(n)
119for m in n.outbound_nodes:
120 G[n]['out'].remove(m)
121 G[m]['in'].remove(n)
122if len(G[m]['in']) == 0 :
123 S.add(m)
124return L
125
126
127
128#######################使⽤⽅法##############################################
129#⾸先由图的定义执⾏顺序
130#graph = topological_sort(feed_dict)
131def forward_and_backward(graph):
132for n in graph :
133 n.forward()
134
135for n in graph[:: -1]:
136 n.backward()
137
138#对各个模块进⾏正向计算和反向求导
139#forward_and_backward(graph)
140
141#########################介绍梯度下降################
142def sgd_update(trainables, learning_rate = 1e-2):
143for t in trainables :
144 t.value = t.value - learning_rate * t.gradients[t]
145
146###########使⽤这个模型#################################
147from sklearn.utils import resample
148from sklearn import datats
149
150# %matplotlib inline
151
152 data = datats.load_iris()
153 X_ = data.data
154 y_ = data.target
155 y_[y_ == 2] = 1 # 0 for virginica, 1 for not virginica
156print(X_.shape, y_.shape) # out (150,4) (150,)
157
158########################⽤写的模块来定义这个神经⽹络#########################
159
160 np.random.ed(0)
161 n_features = X_.shape[1]
162 n_class = 1
163 n_hidden = 3
164
165 X, y = Input(), Input()
166 W1, b1 = Input(), Input()
167 W2, b2 = Input(), Input()
168
169 l1 = Linear(X, W1, b1)
170 s1 = Sigmoid(l1)
171 l2 = Linear(s1, W2, b2)
172 t1 = Sigmoid(l2)
173 cost = MSE(y, t1)
174
175
176###########训练模型###########################################
177#随即初始化参数值
178 W1_0 = np.random.random(X_.shape[1] * n_hidden).reshape([X_.shape[1], n_hidden])
179 W2_0 = np.random.random(n_hidden * n_class).reshape([n_hidden, n_class])
180 b1_0 = np.random.random(n_hidden)
181 b2_0 = np.random.random(n_class)
182
183#将输⼊值带⼊算⼦
184 feed_dict = {
185 X: X_, y: y_,
186 W1:W1_0, b1: b1_0,
187 W2:W2_0, b2: b2_0
188 }
189
190#训练参数
191#这⾥训练100轮(eprochs)，每轮抽4个样本(batch_size)，训练150/4次(steps_per_eproch)，学习率 0.1 192 epochs = 100
193 m = X_.shape[0]
194 batch_size = 4
195 steps_per_eproch = m // batch_size
196 lr = 0.1
197
198 graph = topological_sort(feed_dict)
199 trainables = [W1, b1,W2, b2]
200
201 l_Mat_W1 = [W1_0]
202 l_Mat_W2 = [W2_0]
203
204 l_loss = []
205for i in range(epochs):
206 loss = 0
207for j in range(steps_per_eproch):
208 X_batch, y_batch = resample(X_, y_, n_samples = batch_size)
209 X.value = X_batch
210 y.value = y_batch
211
212 forward_and_backward(graph)
213 sgd_update(trainables, lr)
214 loss += graph[-1].value
215
216 l_loss.append(loss)
217if i % 10 ==9 :
218print("Eproch %d, Loss = %1.5f" % (i, loss))
219
220
221#图形化显⽰
222 plt.plot(l_loss)
223 plt.title("Cross Entropy value")
224 plt.xlabel("Eproch")
225 plt.ylabel("Loss")
226 plt.show()
227
228
229##########最后⽤模型预测所有的数据的情况230 X.value = X_
231 y.value = y_
232for n in graph:
233 n.forward()
234
235
236 plt.plot(graph[-2].value.ravel())
237 plt.title("predict for all 150 Iris data")
238 plt.xlabel("Sample ID")
239 plt.ylabel("Probability for not a virginica") 240 plt.show()
