-
Notifications
You must be signed in to change notification settings - Fork 3
/
py_nn_use.py
315 lines (265 loc) · 10 KB
/
py_nn_use.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
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
# coding:utf-8
# 《python神经网络编程》实操代码
# 具体应用
import numpy as np
import scipy.special
import run
import pandas as pd
import matplotlib.pyplot as plt
import optuna
import optuna.visualization as pv
import cv2
import glob
# 神经网络类
class NN:
def __init__(self, inputnodes, hiddennodes, outputnodes, learningrate):
# 设置输入、隐藏和输出层维度
self.inodes = inputnodes
self.hnodes = hiddennodes
self.onodes = outputnodes
# simple random number
# self.wih = (np.random.rand(self.hnodes, self.inodes) - 0.5)
# self.who = (np.random.rand(self.onodes, self.hnodes) - 0.5)
# Normal distribution
# average = 0
# Standard deviation = 1/evolution of number of nodes passed in
# 用正态分布随机数初始化权重
self.wih = np.random.normal(0, pow(self.hnodes, -0.5), (self.hnodes, self.inodes))
self.who = np.random.normal(0, pow(self.onodes, -0.5), (self.onodes, self.hnodes))
# 学习率
self.lr = learningrate
# 用sigmoid函数做激活函数
self.activation_function = lambda x: scipy.special.expit(x)
# 训练神经网络
def train(self, inputs_list, targets_list):
# 将数据转换为二维数组
inputs = np.array(inputs_list, ndmin=2).T
targets = np.array(targets_list, ndmin=2).T
# 利用传输矩阵wih,计算隐藏层输入
hidden_inputs = np.dot(self.wih, inputs)
# 计算隐藏层输出,激活函数
hidden_outputs = self.activation_function(hidden_inputs)
# 利用传输矩阵who,计算输出层输入
final_inputs = np.dot(self.who, hidden_outputs)
# 用激活函数计算输出信号
final_outputs = self.activation_function(final_inputs)
# 计算误差值
output_errors = targets - final_outputs
# 按权重分配误差
hidden_errors = np.dot(self.who.T, output_errors)
# update the weights for the links between the hidden and output layers
# wj,k = learningrate * error * sigmoid(ok) * (1 - sigmoid(ok)) · oj^T
# 更新隐藏层及输出层之间的权重值
self.who += self.lr * np.dot(
(output_errors * final_outputs * (1.0 - final_outputs)),
np.transpose(hidden_outputs))
# update the weights for the links between the input and hidden layers
# 更新输入层及隐藏层之间的权重值
self.wih += self.lr * np.dot(
(hidden_errors * hidden_outputs * (1.0 - hidden_outputs)),
np.transpose(inputs))
# 前向传播
def query(self, inputs_list):
# 输入矩阵
inputs = np.array(inputs_list, ndmin=2).T
# calculate signals into hidden layer
# 利用传输矩阵wih,计算隐藏层输入
hidden_inputs = np.dot(self.wih, inputs)
# calculate the signals emerging from hidden layer
# 计算隐藏层输出,激活函数
hidden_outputs = self.activation_function(hidden_inputs)
# calculate signals into final output layer
# 利用传输矩阵who,计算输出层输入
final_inputs = np.dot(self.who, hidden_outputs)
# calculate the signals emerging from final output layer
final_outputs = self.activation_function(final_inputs)
return final_outputs
# 加载数据
@run.change_dir
def loadData():
# load the mnist training data CSV file into a list
training_data_file = open("mnist_train.csv", 'r')
training_data_list = training_data_file.readlines()
training_data_file.close()
testing_data_file = open("mnist_test.csv", 'r')
testing_data_list = testing_data_file.readlines()
testing_data_file.close()
return training_data_list, testing_data_list
# 创建模型
def init_model(input_nodes, hidden_nodes, output_nodes, learning_rate):
# create instance of neural network
n = NN(input_nodes, hidden_nodes, output_nodes, learning_rate)
return n
# 训练过程
def train(n, epochs, training_data_list, output_nodes):
# 对训练过程进行循环
for e in range(epochs):
print("第{}轮".format(e))
for record in training_data_list:
# split the record by the ',' commas
# 通过','将数分段
all_values = record.split(',')
# scale and shift the inputs
# 将所有的像素点的值转换为0.01-1.00
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99 + 0.01)
# creat the target output values
# 创建标签输出值
targets = np.zeros(output_nodes) + 0.01
# all_values[0] is the target label for this record
# 10个输出值,对应的为0.99,其他为0.01
targets[int(all_values[0])] = 0.99
# 传入网络进行训练
n.train(inputs, targets)
return n
# 获取预测准确率
def getScores(n, testing_data_list):
# 创建一个空白的计分卡
scorecard = []
# 遍历测试数据
for record in testing_data_list:
all_values = record.split(',')
# 提取正确的标签
correct_label = int(all_values[0])
# print(correct_label, 'correct label')
# 读取像素值并转换
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99 + 0.01)
# 通过神经网络得出结果
outputs = n.query(inputs)
# 结果
label = np.argmax(outputs)
# print(label, "network's answer")
# 标签相同,计分卡加一,否则加零
if (label == correct_label):
scorecard.append(1)
else:
scorecard.append(0)
# 输出计分卡
# print(scorecard)
# 输出分数
scorecard_array = np.asarray(scorecard)
return scorecard_array
# 解MINST手写数字识别问题
@run.change_dir
@run.timethis
def minst(trial):
input_nodes = 784
hidden_nodes = trial.suggest_categorical("hidden_dim", [50, 100, 200, 300])
output_nodes = 10
# 学习率
learning_rate = trial.suggest_discrete_uniform("learning_rate", 0.01, 0.81, 0.1)
n = init_model(input_nodes, hidden_nodes, output_nodes, learning_rate)
training_data_list, testing_data_list = loadData()
# 训练
epochs = trial.suggest_int("epochs:", 1, 10)
n = train(n, epochs, training_data_list, output_nodes)
# 测试
res = getScores(n, testing_data_list)
return res.sum() / res.size
# 画图
@run.change_dir
def draw_results(study):
# 优化历史
plt.figure()
fig = pv.plot_optimization_history(study)
fig.write_image("./output/opt_his.png")
plt.close()
# 等高线图
plt.figure()
fig = pv.plot_contour(study)
fig.write_image("./output/opt_contour.png")
plt.close()
# 经验分布图
plt.figure()
fig = pv.plot_edf(study)
fig.write_image("./output/opt_edf.png")
plt.close()
# 高维参数
plt.figure()
fig = pv.plot_parallel_coordinate(study)
fig.write_image("./output/opt_coordinate.png")
plt.close()
# 手写数字识别应用
# 处理输入数据
@run.change_dir
def data_process():
targets = []
datas = []
for file in glob.glob(r"./pic/*.jpg"):
targets.append(int(file.split("/")[2].split(".")[0]))
if targets[-1] == 10:
targets[-1] = 0
img_array = cv2.imread(file)
img_array = cv2.resize(img_array, (28, 28))
img_array = cv2.cvtColor(img_array, cv2.COLOR_BGR2GRAY)
height,width = img_array.shape
dst = np.zeros((height,width),np.uint8)
for i in range(height):
for j in range(width):
dst[i,j] = 255 - img_array[i,j]
img_array = dst.reshape(784)
datas.append(img_array)
return (targets, datas)
# 训练模型
@run.timethis
def trainModel():
print("开始训练")
input_nodes = 784
hidden_nodes = 300
output_nodes = 10
learning_rate = 0.11
epochs = 8
model = NN(input_nodes, hidden_nodes, output_nodes, learning_rate)
training_data_list, _ = loadData()
# 对训练过程进行循环
for e in range(epochs):
for record in training_data_list:
# 通过','将数分段
all_values = record.split(',')
# 将所有的像素点的值转换为0.01-1.00
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99 + 0.01)
# 创建标签输出值
targets = np.zeros(output_nodes) + 0.01
# 10个输出值,对应的为0.99,其他为0.01
targets[int(all_values[0])] = 0.99
# 传入网络进行训练
model.train(inputs, targets)
return model
# 用模型识别实际数据
def testModel(model, test_datas, targets):
n = len(test_datas)
correct = 0
for i in range(n):
# 用模型得出预测值
outputs = model.query(test_datas[i])
# 转换为结果
label = np.argmax(outputs)
print("预测结果{},实际结果{}".format(label, targets[i]))
if label == targets[i]:
correct += 1
return correct/n
if __name__ == "__main__":
"""
input_nodes = 3
hidden_nodes = 3
output_nodes = 3
learning_rate = 0.3
# n = NN(input_nodes, hidden_nodes, output_nodes, learning_rate)
# print(n.query([1.0, 0.5, -0.5]))
# minst()
study = optuna.create_study(direction="maximize")
study.optimize(minst, n_trials=100)
print("结果:", study.best_params)
print(study.best_value)
print(study.best_trial)
if pv.is_available:
print("结果作图")
draw_results(study)
else:
print("不能作图")
"""
# 具体应用模型
# 目前得到的最佳参数:{'hidden_dim': 300, 'learning_rate': 0.11, 'epochs:': 9}
targets, datas = data_process()
model = trainModel()
score = testModel(model, datas, targets)
print("模型预测准确率:{}".format(score))