一个bp神经网络的测试函数用python

一个BP神经网络的测试函数可以用Python编写，如下所示：

import numpy as np

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

def sigmoid_derivative(x):
    return x * (1 - x)

def train(inputs, outputs, epochs, learning_rate):
    input_layer_size = inputs.shape
    hidden_layer_size = 4
    output_layer_size = 1

    weights_hidden = np.random.uniform(size=(input_layer_size, hidden_layer_size))
    weights_output = np.random.uniform(size=(hidden_layer_size, output_layer_size))

    for epoch in range(epochs):
        # Forward pass
        hidden_layer_input = np.dot(inputs, weights_hidden)
        hidden_layer_output = sigmoid(hidden_layer_input)
        output_layer_input = np.dot(hidden_layer_output, weights_output)
        output_layer_output = sigmoid(output_layer_input)

        # Backward pass
        output_error = outputs - output_layer_output
        output_delta = output_error * sigmoid_derivative(output_layer_output)

        hidden_error = np.dot(output_delta, weights_output.T)
        hidden_delta = hidden_error * sigmoid_derivative(hidden_layer_output)

        # Update weights
        weights_output += np.dot(hidden_layer_output.T, output_delta) * learning_rate
        weights_hidden += np.dot(inputs.T, hidden_delta) * learning_rate

    return weights_hidden, weights_output

def predict(inputs, weights_hidden, weights_output):
    hidden_layer_input = np.dot(inputs, weights_hidden)
    hidden_layer_output = sigmoid(hidden_layer_input)
    output_layer_input = np.dot(hidden_layer_output, weights_output)
    output_layer_output = sigmoid(output_layer_input)
    return output_layer_output

# Example usage:
inputs = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
outputs = np.array([, , , ])

weights_hidden, weights_output = train(inputs, outputs, epochs=10000, learning_rate=0.1)
predictions = predict(inputs, weights_hidden, weights_output)
print(predictions)

这个测试函数包含了一个简单的BP神经网络的训练和预测过程。首先定义了激活函数（sigmoid）及其导数，然后实现了训练函数（train）和预测函数（predict）。在示例用法中，我们创建了一个简单的异或（XOR）问题的输入和输出数据，并使用训练好的神经网络进行预测。