Figure 7
The table shows two column headers labeled “Parameter” and “Value” at the top, followed by grouped sections separated by horizontal lines, with entries listed from top to bottom. Under “Conv-1 D Layer”, Row 1: Parameter: Number of Filters. Value: 32. Row 2: Parameter: Activation Function. Value: ReLU. Row 3: Parameter: Padding. Value: Same. Row 4: Parameter: Regularization. Value: L 2 (0.001). Row 5: Parameter: Max Pooling Pool Size. Value: 2. Under “Inception Module”, Row 1: Parameter: Conv1 D Filters. Value: 32. Row 2: Parameter: Conv1 D Kernel Sizes. Value: 1, 3, 5. Row 3: Parameter: Concatenation. Value: Yes. Row 4: Parameter: Max Pooling Pool Size. Value: 2. Row 5: Parameter: Dropout Rate. Value: 30 percent. Under “L S T M Layer”, Row 1: Parameter: Number of Units. Value: 256. Row 2: Parameter: Return Sequences. Value: False. Row 3: Parameter: Regularization. Value: L 2 (0.001). Under “Dense Layer”, Row 1: Parameter: Number of Units. Value: 256. Row 2: Parameter: Activation Function. Value: ReLU. Row 3: Parameter: Regularization. Value: L 2 (0.001). Row 4: Parameter: Batch Normalization. Value: Yes. Under “Output Layer”, Row 1: Parameter: Number of Units. Value: 1. Row 2: Parameter: Activation Function. Value: Sigmoid. Under “Training Hyperparameters”, Row 1: Parameter: Optimizer Type. Value: Adam. Row 2: Parameter: Learning Rate. Value: 5 e negative 4. Row 3: Parameter: Patience. Value: 20 epochs. Row 4: Parameter: Restore Best Weights. Value: Yes. Row 5: Parameter: Batch Size. Value: 64. Row 6: Parameter: Epochs. Value: 200.The proposed CNN-LSTM hybrid model hyperparameters. Source: Created by the authors