《python神经网络编程 塔里克》代码笔记

花了三天读了塔里克的python神经网络编程，是一本真正的神经网络初学者书籍，让我学会了最简单的三层神经网络编程，强烈安利给大伙，下面发注释代码记录一下。

import numpy # 数值计算库
import scipy.special # 提供sigmoid函数的库
import matplotlib.pyplot # 绘图库
class neuralNetwork:
    def __init__(self, inputnodes, hiddennodes, outputnodes, learningrate):
        # 初始化神经网络，需要有发射层，隐藏层，输出层
        self.inodes = inputnodes
        self.hnodes = hiddennodes
        self.onodes = outputnodes
        # 使用正态分布取权重矩阵，参数：均值，标准差，矩阵大小 wih，who分别是输入层和输出层对于隐藏层的权重
        # 这里要注意矩阵创建的时候是几乘几的，不然下面做矩阵乘法会出错
        # 还有学习率和激活函数的创建
        self.wih = numpy.random.normal(0, pow(self.hnodes, -0.5), (self.hnodes, self.inodes))
        self.who = numpy.random.normal(0 ,pow(self.hnodes, -0.5), (self.onodes, self.hnodes))
        self.lr = learningrate
        self.activation_function = lambda x: scipy.special.expit(x)
   
    def train(self, inputs_list, targets_list):
        # 训练并对权重矩阵进行更新，更新公式书上有一步一步推导，最后两行是核心，一定要理解
        inputs = numpy.array(inputs_list, ndmin=2).T
        targets = numpy.array(targets_list, ndmin=2).T
        hidden_inputs = numpy.dot(self.wih, inputs)
        hidden_outputs = self.activation_function(hidden_inputs)
        final_inputs = numpy.dot(self.who, hidden_outputs)
        final_outputs = self.activation_function(final_inputs)
        output_errors = targets - final_outputs
        hidden_errors = numpy.dot(self.who.T, output_errors)
        self.who += self.lr * numpy.dot((output_errors * final_outputs * (1 - final_outputs)), numpy.transpose(hidden_outputs))
        self.wih += self.lr * numpy.dot((hidden_errors * hidden_outputs * (1 - hidden_outputs)), numpy.transpose(inputs))
        
        
    def query(self, inputs_list):
        # 测试，简单矩阵计算得出结果
        inputs = numpy.array(inputs_list, ndmin=2).T
        hidden_inputs = numpy.dot(self.wih, inputs)
        hidden_outputs = self.activation_function(hidden_inputs)
        final_inputs = numpy.dot(self.who, hidden_outputs)
        final_outputs = self.activation_function(final_inputs)
        return final_outputs

input_nodes = 784  # 为什么784，因为作者给的训练集重一个图片对应785个数字，第一个数字是答案本身，所以输入只要输入784
hidden_nodes = 100 # 为什么隐藏层是100个节点？ 比输入层小，比输出层大
output_nodes = 10
learning_rate = 0.3
n = neuralNetwork(input_nodes, hidden_nodes, output_nodes, learning_rate) # 创建了！
training_data_file = open("C:/Users/15186/Desktop/mnist_dataset/mnist_train_100.csv",'r')
training_data_list = training_data_file.readlines()
training_data_file.close()
def training():  # 训练出一个合适的神经网络，改变网络的权重
    for record in training_data_list:
        all_values = record.split(',')
        inputs = (numpy.asfarray(all_values[1:])/255.0 * 0.99) + 0.01 # asfarray将原本是字符的列表转换成整数列表，并控制数在0.1到0.99
        targets = numpy.zeros(output_nodes) + 0.01
        targets[int(all_values[0])] = 0.99 # 给定目标集正确答案
        n.train(inputs, targets)

epochs = 100 # 训练回合数
for _ in range(epochs):
    training()
print("训练完成") 

test_data_file = open("C:/Users/15186/Desktop/mnist_dataset/mnist_test_10.csv",'r')
test_data_list =test_data_file.readlines() #读取数据
test_data_file.close

scorecard = [] # 积分列表
for record in test_data_list:
    all_values = record.split(',')
    correct_label = int(all_values[0])
    inputs = (numpy.asfarray(all_values[1:])/225.0*0.99)+0.01
    outputs = n.query(inputs) # 测试
    label = numpy.argmax(outputs) # 取最大数的标签
    if(label == correct_label):
        scorecard.append(1)
    else:
        scorecard.append(0)
        pass
print(scorecard)