手写数字的数据集MNIST的一个小程序和试验结果分析

import numpy as np
import matplotlib.pyplot as plt

import scipy.special
import datetime

class neuralNetwork:
    def __init__(self, inputnodes, hiddennodes, outputnodes,learningrate):
        self.inodes = inputnodes
        self.hnodes = hiddennodes
        self.onodes = outputnodes
        self.with1 = np.random.normal(0.0,pow(self.hnodes,-0.5),(self.hnodes,self.inodes))
        self.who = np.random.normal(0.0,pow(self.onodes,-0.5),(self.onodes,self.hnodes))
        self.lr = learningrate
 
        self.activation_function = lambda x: scipy.special.expit(x)
        pass
    def train(self,inputs_list,targets_list):
        inputs = np.array(inputs_list,ndmin=2).T
        targets = np.array(targets_list,ndmin=2).T
        hidden_inputs = np.dot(self.with1,inputs)
        hidden_outputs = self.activation_function(hidden_inputs)
        
        final_inputs = np.dot(self.who,hidden_outputs)
        final_outputs = self.activation_function(final_inputs)

        output_errors = targets - final_outputs
        hidden_errors = np.dot(self.who.T,output_errors)

        self.who += self.lr*np.dot((output_errors*final_outputs*(1-final_outputs)),np.transpose(hidden_outputs)) 
        
        self.with1 += self.lr*np.dot((hidden_errors*hidden_outputs*(1-hidden_outputs)),np.transpose(inputs))
        pass

    def query(self,inputs_list):
        inputs = np.array(inputs_list,ndmin=2).T
        
        hidden_inputs = np.dot(self.with1,inputs)
        hidden_outputs = self.activation_function(hidden_inputs)
        
        final_inputs = np.dot(self.who,hidden_outputs)
        final_outputs = self.activation_function(final_inputs)

        return final_outputs

start1 = datetime.datetime.now()

input_nodes = 784
hidden_nodes = 200 #200
output_nodes = 10
learning_rate = 0.1

n = neuralNetwork(input_nodes,hidden_nodes,output_nodes,learning_rate)

training_data_file = open("mnist_dataset/mnist_train.csv",'r')
training_data_list = training_data_file.readlines()
training_data_file.close()

epochs = 5 # 权重更新循环的次数

for e in range(epochs):
    for record in training_data_list:
        all_values = record.split(',')
        inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99) + 0.01
        targets = np.zeros(output_nodes) +0.01
        targets[int(all_values[0])] = 0.99
        n.train(inputs,targets)
        pass
    pass

end1 = datetime.datetime.now()
print("程序训练所需时间为：" + str((end1 - start1).seconds) + "秒")
 
test_data_file = open("mnist_dataset/mnist_test.csv",'r')
test_data_list = test_data_file.readlines()
test_data_file.close()

start2 = datetime.datetime.now()

scorecared = []
for record in test_data_list:
    all_values = record.split(',')
    correct_lable = int(all_values[0])
    inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99) + 0.01
    outputs = n.query(inputs)
    label = np.argmax(outputs)
    if (label == correct_lable):
        scorecared.append(1)
    else:
        scorecared.append(0)
        pass
    pass

scorecared_array = np.asarray(scorecared)
print ("performance = ", scorecared_array.sum() / scorecared_array.size)
end2 = datetime.datetime.now()
print("程序测试所需时间为：" + str((end2 - start2).seconds) + "秒")
# print("程序所需时间为：" + str((end2 - start1).seconds) + "秒")

此代码来自于《Python圣经网络编程》作者塔里克.拉希德著  林赐 译

此程序中使用的数据来自，手写数字的数据集MNIST，

根据作者所述进行试验，对权重更新的次数（作者建议为5）、学习率（作者建议为0.1）、隐藏层节点的数目（作者建议为200）进行试验分析，试验结果统计如下：

 

 

 

 

这些试验数据只供参考，由于能力有限，有不足之处，还望多提建议。谢谢，，，，，，