误差修正模型python代码

以下是一个简单的误差修正模型的 Python 代码示例：

import numpy as np

# 定义输入数据
X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
# 定义期望输出
Y = np.array([[0], [1], [1], [0]])

# 定义权重和偏置
w1 = np.random.randn(2, 4)
b1 = np.zeros((1, 4))
w2 = np.random.randn(4, 1)
b2 = np.zeros((1, 1))

# 定义学习率和迭代次数
lr = 0.1
epochs = 10000

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

# 训练模型
for i in range(epochs):
    # 前向传播
    z1 = np.dot(X, w1) + b1
    a1 = sigmoid(z1)
    z2 = np.dot(a1, w2) + b2
    y_pred = sigmoid(z2)

    # 计算损失函数
    loss = np.square(Y - y_pred).mean()

    # 反向传播
    dloss_dy_pred = -2 * (Y - y_pred)
    dy_pred_dz2 = sigmoid(z2) * (1 - sigmoid(z2))
    dz2_da1 = w2.T
    da1_dz1 = sigmoid(z1) * (1 - sigmoid(z1))
    dz1_dw1 = X.T

    dloss_dz2 = dloss_dy_pred * dy_pred_dz2
    dloss_da1 = np.dot(dloss_dz2, dz2_da1.T)
    dloss_dz1 = dloss_da1 * da1_dz1
    dloss_dw1 = np.dot(dz1_dw1, dloss_dz1)
    dloss_db1 = dloss_dz1.sum(axis=0, keepdims=True)
    dloss_dw2 = np.dot(a1.T, dloss_dz2)
    dloss_db2 = dloss_dz2.sum(axis=0, keepdims=True)

    # 更新权重和偏置
    w1 -= lr * dloss_dw1
    b1 -= lr * dloss_db1
    w2 -= lr * dloss_dw2
    b2 -= lr * dloss_db2

    # 打印损失函数
    if i % 1000 == 0:
        print(f"Epoch {i}, loss: {loss:.4f}")

# 预测新数据
new_data = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
new_pred = sigmoid(np.dot(sigmoid(np.dot(new_data, w1) + b1), w2) + b2)
print(new_pred)