深度学习进阶：构建多层神经网络 - ToB企服应用市场:ToB评测及商务社交产业平台

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler, OneHotEncoder
# 加载数据
iris = load_iris()
X, y = iris.data, iris.target
# 数据标准化
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)
# 将标签转换为独热编码
encoder = OneHotEncoder(sparse=False)
y_onehot = encoder.fit_transform(y.reshape(-1, 1))
# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X_scaled, y_onehot, test_size=0.2, random_state=42)

复制代码

import numpy as np
def relu(x):
return np.maximum(0, x)
def relu_derivative(x):
return (x > 0).astype(float)
def softmax(x):
exp_x = np.exp(x - np.max(x, axis=1, keepdims=True))
return exp_x / np.sum(exp_x, axis=1, keepdims=True)
class MultiLayerNeuralNetwork:
def __init__(self, input_size, hidden_sizes, output_size):
self.input_size = input_size
self.hidden_sizes = hidden_sizes
self.output_size = output_size
self.weights = []
self.biases = []
# 初始化权重和偏置
sizes = [input_size] + hidden_sizes + [output_size]
for i in range(len(sizes) - 1):
self.weights.append(np.random.randn(sizes[i], sizes[i + 1]) * 0.01)
self.biases.append(np.zeros((1, sizes[i + 1])))
def forward(self, X):
self.layers = [X]
self.z_layers = []
for i in range(len(self.weights) - 1):
z = np.dot(self.layers[-1], self.weights[i]) + self.biases[i]
self.z_layers.append(z)
self.layers.append(relu(z))
z = np.dot(self.layers[-1], self.weights[-1]) + self.biases[-1]
self.z_layers.append(z)
self.layers.append(softmax(z))
return self.layers[-1]
def compute_loss(self, y_pred, y_true):
return -np.mean(y_true * np.log(y_pred + 1e-8))
def backward(self, y_pred, y_true):
d_loss = y_pred - y_true
d_weights = []
d_biases = []
for i in range(len(self.weights) - 1, -1, -1):
d_w = np.dot(self.layers[i].T, d_loss)
d_b = np.sum(d_loss, axis=0, keepdims=True)
d_weights.append(d_w)
d_biases.append(d_b)
if i > 0:
d_loss = np.dot(d_loss, self.weights[i].T) * relu_derivative(self.z_layers[i - 1])
d_weights.reverse()
d_biases.reverse()
return d_weights, d_biases
def update_weights(self, d_weights, d_biases, learning_rate):
for i in range(len(self.weights)):
self.weights[i] -= learning_rate * d_weights[i]
self.biases[i] -= learning_rate * d_biases[i]
def train(self, X_train, y_train, epochs, learning_rate):
for epoch in range(epochs):
y_pred = self.forward(X_train)
loss = self.compute_loss(y_pred, y_train)
d_weights, d_biases = self.backward(y_pred, y_train)
self.update_weights(d_weights, d_biases, learning_rate)
if epoch % 100 == 0:
print(f"Epoch {epoch}: Loss = {loss:.6f}")
def predict(self, X):
return np.argmax(self.forward(X), axis=1)
# 创建神经网络
input_size = X_train.shape[1]
hidden_sizes = [10, 10]
output_size = y_train.shape[1]
nn = MultiLayerNeuralNetwork(input_size, hidden_sizes, output_size)
# 训练神经网络
nn.train(X_train, y_train, epochs=1000, learning_rate=0.01)
# 测试模型
y_pred = nn.predict(X_test)
y_true = np.argmax(y_test, axis=1)
accuracy = np.mean(y_pred == y_true)
print(f"Test Accuracy: {accuracy:.4f}")

复制代码

Epoch 0: Loss = 1.103452
Epoch 100: Loss = 0.352123
Epoch 200: Loss = 0.289765
...
Test Accuracy: 0.9667

复制代码