bp神经网络python代码

以下是一个简单的 Python 代码实现 BP 神经网络的例子：

import numpy as np

class NeuralNetwork:
    def __init__(self, layers_size):
        self.layers_size = layers_size
        self.weights = []
        self.bias = []
        self.num_layers = len(layers_size)

        for i in range(self.num_layers - 1):
            self.weights.append(np.random.rand(layers_size[i], layers_size[i+1]))
            self.bias.append(np.random.rand(layers_size[i+1]))

    def sigmoid(self, x):
        return 1/(1+np.exp(-x))

    def feedforward(self, X):
        output = X
        for i in range(self.num_layers - 2):
            output = self.sigmoid(np.dot(output, self.weights[i]) + self.bias[i])
        output = np.dot(output, self.weights[-1]) + self.bias[-1]
        return output

    def backpropagation(self, X, y, learning_rate):
        outputs = []
        output = X
        for i in range(self.num_layers - 2):
            output = self.sigmoid(np.dot(output, self.weights[i]) + self.bias[i])
            outputs.append(output)
        output = np.dot(output, self.weights[-1]) + self.bias[-1]
        outputs.append(output)

        deltas = [None] * self.num_layers
        deltas[-1] = y - outputs[-1]
        for i in reversed(range(self.num_layers - 1)):
            error = deltas[i].dot(self.weights[i].T)
            deltas[i-1] = error * outputs[i-1] * (1 - outputs[i-1])

        for i in range(self.num_layers - 1):
            self.weights[i] += learning_rate * outputs[i].T.dot(deltas[i])
            self.bias[i] += learning_rate * deltas[i]

    def train(self, X, y, learning_rate=0.1, epochs=1000):
        for i in range(epochs):
            self.backpropagation(X, y, learning_rate)

    def predict(self, X):
        return self.feedforward(X)

以上代码实现了一个三层神经网络，可以根据输入的数据进行训练和预测。其中 `layers_size` 参数是一个数组，指定了每层神经元的数量。 `backpropagation` 方法实现了反向传播算法， `train` 方法可以用来训练模型， `predict` 方法用来进行预测。