import tensorflow as tf from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense, InputLayer, Dropout, Conv1D, Flatten, Reshape, MaxPooling1D from tensorflow.keras.optimizers.legacy import Adam from sklearn.model_selection import KFold import numpy as np class_weights = {0: 1.0, 1: 2.0, 2: 1.5} # Assign weights to each class model.fit(X_train, y_train, class_weight=class_weights) from tensorflow.keras.losses import CategoricalCrossentropy loss = CategoricalCrossentropy(label_smoothing=0.1) # Apply smoothing to reduce overconfidence model.compile(loss=loss, optimizer=opt, metrics=['accuracy']) from tensorflow.keras.losses import CategoricalCrossentropy loss = CategoricalCrossentropy(label_smoothing=0.1) # Apply smoothing to reduce overconfidence model.compile(loss=loss, optimizer=opt, metrics=['accuracy']) # Parameters EPOCHS = args.epochs or 200 LEARNING_RATE = args.learning_rate or 0.005 BATCH_SIZE = args.batch_size or 32 k = 5 # Number of folds # Dataset as NumPy arrays (ensure you load these correctly) X = np.array(X_train) # Features y = np.array(Y_train) # Labels (one-hot encoded) # k-Fold Cross-Validation kfold = KFold(n_splits=k, shuffle=True, random_state=42) fold_accuracies = [] for fold, (train_index, val_index) in enumerate(kfold.split(X)): print(f"\n--- Fold {fold + 1} ---") # Split the data for the current fold X_train_fold, X_val_fold = X[train_index], X[val_index] y_train_fold, y_val_fold = y[train_index], y[val_index] # Create a new model for each fold model = Sequential() model.add(Reshape((int(input_length / 40), 40), input_shape=(input_length,))) model.add(Conv1D(8, kernel_size=3, padding='same', activation='relu')) model.add(MaxPooling1D(pool_size=2, strides=2, padding='same')) model.add(Dropout(0.25)) model.add(Conv1D(16, kernel_size=3, padding='same', activation='relu')) model.add(MaxPooling1D(pool_size=2, strides=2, padding='same')) model.add(Dropout(0.25)) model.add(Flatten()) model.add(Dense(classes, name='y_pred', activation='softmax')) # Optimizer opt = Adam(learning_rate=LEARNING_RATE, beta_1=0.9, beta_2=0.999) # Compile the model model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy']) # Train the model history = model.fit( X_train_fold, y_train_fold, epochs=EPOCHS, batch_size=BATCH_SIZE, validation_data=(X_val_fold, y_val_fold), verbose=1 ) # Evaluate the model on the validation fold val_loss, val_accuracy = model.evaluate(X_val_fold, y_val_fold, verbose=0) fold_accuracies.append(val_accuracy) print(f"Validation accuracy for fold {fold + 1}: {val_accuracy:.4f}") # Average accuracy across all folds mean_accuracy = np.mean(fold_accuracies) print(f"\nMean accuracy across {k} folds: {mean_accuracy:.4f}")