多层感知机python实现不调库

多层感知机（Multilayer Perceptron, MLP）是一种常用的人工神经网络模型，用于解决分类和回归问题。下面是一个使用Python实现多层感知机的示例代码，不依赖任何外部库：

import numpy as np

class MLP:
    def __init__(self, num_inputs, num_hidden, num_outputs):
        self.num_inputs = num_inputs
        self.num_hidden = num_hidden
        self.num_outputs = num_outputs
        
        # 初始化权重和偏置
        self.weights1 = np.random.randn(self.num_inputs, self.num_hidden)
        self.bias1 = np.zeros((1, self.num_hidden))
        self.weights2 = np.random.randn(self.num_hidden, self.num_outputs)
        self.bias2 = np.zeros((1, self.num_outputs))
    
    def sigmoid(self, x):
        return 1 / (1 + np.exp(-x))
    
    def forward(self, X):
        # 输入层到隐藏层
        self.hidden_output = np.dot(X, self.weights1) + self.bias1
        self.hidden_activation = self.sigmoid(self.hidden_output)
        
        # 隐藏层到输出层
        self.output = np.dot(self.hidden_activation, self.weights2) + self.bias2
        
        return self.sigmoid(self.output)
    
    def backward(self, X, y, learning_rate):
        # 反向传播
        output_error = y - self.forward(X)
        output_delta = output_error * (self.sigmoid(self.output) * (1 - self.sigmoid(self.output)))
        
        hidden_error = np.dot(output_delta, self.weights2.T)
        hidden_delta = hidden_error * (self.sigmoid(self.hidden_output) * (1 - self.sigmoid(self.hidden_output)))
        
        # 更新权重和偏置
        self.weights2 += learning_rate * np.dot(self.hidden_activation.T, output_delta)
        self.bias2 += learning_rate * np.sum(output_delta, axis=0, keepdims=True)
        self.weights1 += learning_rate * np.dot(X.T, hidden_delta)
        self.bias1 += learning_rate * np.sum(hidden_delta, axis=0)
    
    def train(self, X, y, num_epochs, learning_rate):
        for epoch in range(num_epochs):
            self.forward(X)
            self.backward(X, y, learning_rate)
        
    def predict(self, X):
        return np.round(self.forward(X))

# 示例用法
X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
y = np.array([[0], [1], [1], [0]])

mlp = MLP(2, 4, 1)
mlp.train(X, y, num_epochs=10000, learning_rate=0.1)

print(mlp.predict(X))

上述代码使用numpy实现了一个简单的多层感知机，包括前向传播、反向传播和训练过程。你可以根据需要调整网络的层数、每层的神经元数量、训练轮数和学习率等参数。这只是一个简单的示例，实际应用中可能需要更复杂的网络结构和调优方法。