网站建设服务器配置施工企业项目负责人现场带班时间
知识点回顾：
- 三种不同的模型可视化方法：推荐torchinfo打印summary+权重分布可视化
- 进度条功能：手动和自动写法，让打印结果更加美观
- 推理的写法：评估模式
作业：调整模型定义时的超参数，对比下效果。
学习率（Learning Rate）
from keras.optimizers import Adam
optimizer = Adam(learning_rate=0.001) # 初始学习率示例正则化参数
model.add(Dense(64, activation='relu', kernel_regularizer=l2(0.01)))
model.add(Dropout(0.3))网格搜索（Grid Search）
from sklearn.model_selection import GridSearchCV
param_grid = {'C': [0.1, 1, 10], 'gamma': [0.01, 0.1]}
grid_search = GridSearchCV(SVC(), param_grid, cv=5)
grid_search.fit(X_train, y_train)随机搜索（Random Search）
from sklearn.model_selection import RandomizedSearchCV
param_dist = {'n_estimators': randint(50, 200), 'max_depth': randint(3, 10)}
random_search = RandomizedSearchCV(RandomForestClassifier(), param_dist, n_iter=10)
random_search.fit(X, y)贝叶斯优化
import optuna
def objective(trial):lr = trial.suggest_float('lr', 1e-5, 1e-1, log=True)batch_size = trial.suggest_categorical('batch_size', [32, 64, 128])return evaluate_model(lr, batch_size)study = optuna.create_study(direction='maximize')
study.optimize(objective, n_trials=50)使用交叉验证确保结果可靠性：
from sklearn.model_selection import cross_val_score
scores = cross_val_score(model, X, y, cv=5, scoring='accuracy')@浙大疏锦行