基于《Python神经网络编程》一书写的代码

# Python:3层神经网络 # 初始代码源于：《Python神经网络编程》一书 # @Author: Hyde Kong # -*- coding: UTF-8 -*- import numpy import scipy.special # 神经网络类定义 class NeuralNetwork: # 初始化神经网络 def __init__(self, inputnodes, hiddennodes, outputnodes, learningrate): # 设置每次输入层、隐藏层、输出层的节点数字 self.inodes = inputnodes self.hnodes = hiddennodes self.onodes = outputnodes # 链接权重矩阵 # 矩阵内的权重设置为w_i_j,意思是从节点i链接到下一层节点j # w11 w21 # w21 w22等 # 用下一层的节点数的开发作为标准方差来初始化权重 self.wih = numpy.random.normal(0.0, pow(self.hnodes, -0.5), (self.hnodes, self.inodes)) self.who = numpy.random.normal(0.0, pow(self.onodes, -0.5), (self.onodes, self.hnodes)) # 学习率 self.lr = learningrate # 激活sgmoid抑制函数 self.activation_function = lambda x: scipy.special.expit(x) pass # 训练神经网络 def train(self,inputs_list,targets_list): # 矩阵转置 inputs = numpy.array(inputs_list, ndmin=2).T targets = numpy.array(targets_list,ndmin=2).T # 计算隐藏层输入信号 hidden_inputs = numpy.dot(self.wih, inputs) # 计算隐藏层输出信号 hidden_outputs = self.activation_function(hidden_inputs) # 计算最终层输入信号 final_inputs = numpy.dot(self.who, hidden_outputs) # 计算最终层输出信号 final_outputs = self.activation_function(final_inputs) # 输出层误差=目标值-实际值 output_errors=targets-final_outputs # 隐藏层误差=根据所连接的权重分割误差，为每个隐藏层节点重组这些误差 hidden_errors=numpy.dot(self.who.T, output_errors) # 更新隐藏层和最终层之间的权重 self.who += self.lr * numpy.dot((output_errors*final_outputs * (1.0 - final_outputs)), numpy.transpose(hidden_outputs)) # 更新输入层和隐藏层之间的权重 self.wih += self.lr * numpy.dot((hidden_errors * hidden_outputs * (1.0 - hidden_outputs)), numpy.transpose(inputs)) pass # 查询神经网络 def query(self, inputs_list): # 矩阵转置 inputs = numpy.array(inputs_list, ndmin=2).T # 计算隐藏层输入信号 hidden_inputs = numpy.dot(self.wih, inputs) # 计算隐藏层输出信号 hidden_outputs = self.activation_function(hidden_inputs) # 计算最终层输入信号 final_inputs = numpy.dot(self.who, hidden_outputs) # 计算最终层输出信号 final_outputs = self.activation_function(final_inputs) return final_outputs