Raw ECG Data: The starting point, represented as a signal array.

ECGDataset: Transforms raw signals into PyTorch Tensors and reshapes with .unsqueeze(1).

DataLoader: Groups tensors into batches of 128 and shuffles data.

Layered Convolutions: 1D Conv layers with dilation (d=1, d=2) and kernel sizes (k=7, k=5).

Normalization: ReLU + BatchNorm for stability.

Global Pooling: Adaptive Avg Pooling to 128 units.

Output Head: Linear layer to 5 AAMI logits.

Forward Pass: Calculate Cross-Entropy Loss.

Backward Pass: Compute gradients with loss.backward().

Optimization: Update weights with Adam Optimizer.

Logging: Track average loss over epochs.