手撕BP（简单版）

本文仅使用numpy 库实现BP神经网络，便于理解其中的反向传播过程以及梯度等。

import numpy as np
import matplotlib.pyplot as plt
class NeuralNetwork:
    def __init__(self, input_size, hidden_size, output_size):
        self.input_size = input_size
        self.hidden_size = hidden_size
        self.output_size = output_size

        self.weights1 = np.random.randn(self.input_size, self.hidden_size)
        self.weights2 = np.random.randn(self.hidden_size, self.output_size)

        self.bias1 = np.zeros((1, self.hidden_size))
        self.bias2 = np.zeros((1, self.output_size))

    def forward(self, X):
        self.z1 = np.dot(X, self.weights1) + self.bias1
        self.a1 = self.sigmoid(self.z1)
        self.z2 = np.dot(self.a1, self.weights2) + self.bias2
        self.a2 = self.sigmoid(self.z2)
        return self.a2

    def backward(self, X, y, output):
        self.error = output - y
        self.delta2 = self.error * self.sigmoid_derivative(self.z2)
        self.delta1 = np.dot(self.delta2, self.weights2.T) * self.sigmoid_derivative(self.z1)

        self.weights2 -= lr * np.dot(self.a1.T, self.delta2)
        self.weights1 -= lr * np.dot(X.T, self.delta1)

        self.bias2 -= lr * np.sum(self.delta2, axis=0)
        self.bias1 -= lr * np.sum(self.delta1, axis=0)

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

    def sigmoid_derivative(self, x):
        return self.sigmoid(x) * (1 - self.sigmoid(x))

X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
y = np.array([[0], [1], [1], [0]])
nn = NeuralNetwork(input_size=2, hidden_size=3, output_size=1)
losses=[]
lr = 0.1
epochs = 10000
for i in range(epochs):
    output = nn.forward(X)
    nn.backward(X, y, output)
    loss = np.mean(np.square(output - y))
    losses.append(loss)
    if i % 100 == 0:
        print(f'epoch:{i}, loss:{loss}')
plt.plot(losses)
plt.show()
predictions = nn.forward(X)
print(predictions)