帮我生成免疫遗传算法优化BP神经网络的代码

当然可以！以下是一个简单的示例代码，展示了如何使用免疫遗传算法优化BP神经网络的权重和偏置：

import numpy as np

# 定义BP神经网络结构和参数
input_size = 4
hidden_size = 5
output_size = 3
learning_rate = 0.1

# 定义适应度函数（均方误差）
def fitness_function(weights):
    # 初始化权重和偏置
    W1 = np.reshape(weights[:input_size * hidden_size], (input_size, hidden_size))
    b1 = np.reshape(weights[input_size * hidden_size:input_size * hidden_size + hidden_size], (hidden_size,))
    W2 = np.reshape(weights[input_size * hidden_size + hidden_size:input_size * hidden_size + hidden_size + hidden_size * output_size], (hidden_size, output_size))
    b2 = np.reshape(weights[input_size * hidden_size + hidden_size + hidden_size * output_size:], (output_size,))

    # 前向传播
    z1 = np.dot(input_data, W1) + b1
    a1 = sigmoid(z1)
    z2 = np.dot(a1, W2) + b2
    a2 = sigmoid(z2)

    # 计算均方误差
    error = np.mean((a2 - output_data) ** 2)

    return 1 / error

# 免疫遗传算法优化
def immune_genetic_algorithm(population_size, chromosome_length, max_generations):
    population = np.random.uniform(-1, 1, (population_size, chromosome_length))
    best_fitness = 0
    best_individual = None

    for generation in range(max_generations):
        fitness_values = np.zeros(population_size)

        # 计算适应度值
        for i in range(population_size):
            fitness_values[i] = fitness_function(population[i])

            # 更新最佳个体
            if fitness_values[i] > best_fitness:
                best_fitness = fitness_values[i]
                best_individual = population[i]

        # 免疫选择
        selected_population = []
        for i in range(population_size):
            if np.random.rand() < fitness_values[i] / np.sum(fitness_values):
                selected_population.append(population[i])
        selected_population = np.array(selected_population)

        # 免疫克隆
        cloned_population = []
        for individual in selected_population:
            clone_size = int(np.round(fitness_function(individual) / np.sum(fitness_values)) * population_size)
            for _ in range(clone_size):
                cloned_population.append(individual)
        cloned_population = np.array(cloned_population)

        # 免疫变异
        mutated_population = []
        for individual in cloned_population:
            mutated_individual = individual + np.random.uniform(-0.1, 0.1, chromosome_length)
            mutated_population.append(mutated_individual)
        mutated_population = np.array(mutated_population)

        # 更新种群
        population[:selected_population.shape[0]] = selected_population
        population[selected_population.shape[0]:] = mutated_population[:population_size - selected_population.shape[0]]

    return best_individual

# 测试数据
input_data = np.array([[0, 0, 1, 1], [0, 1, 0, 1], [1, 1, 0, 0]])
output_data = np.array([[0, 1, 1], [1, 0, 1], [1, 1, 0]])

# 运行免疫遗传算法优化BP神经网络
best_individual = immune_genetic_algorithm(population_size=50, chromosome_length=(input_size * hidden_size) + hidden_size + (hidden_size * output_size) + output_size, max_generations=100)

# 提取最佳个体的权重和偏置
best_weights = best_individual[:input_size * hidden_size + hidden_size + hidden_size * output_size + output_size]
W1 = np.reshape(best_weights[:input_size * hidden_size], (input_size, hidden_size))
b1 = np.reshape(best_weights[input_size * hidden_size:input_size * hidden_size + hidden_size], (hidden_size,))
W2 = np.reshape(best_weights[input_size * hidden_size + hidden_size:input_size * hidden_size + hidden_size + hidden_size * output_size], (hidden_size, output_size))
b2 = np.reshape(best_weights[input_size * hidden_size + hidden_size + hidden_size * output_size:], (output_size,))

请注意，这只是一个简单的示例代码，具体的实现方式可能因应用场景而有所差异。你可以根据自己的需求进行进一步调整和扩展。