Table 5.

Our hyperparameters tuning of the proposed optimized Resnet network.

Category	Element	Description
Network	Conv1D	Increased filter sizes from (64,128,128) to (256, 512, 512) for all Resnet blocks
	Conv1D	Increased kernel sizes from (8,5,3) to (8,7,7) in all Resnet blocks
Training	Labels	Smoothed labels with α = 0.1
	Batch size	Increased training batch size to 256
	Dropout	Introduced Dropout with rate = 0.5
	Epochs	Set training epochs to 120

We found these values to yield the best accuracy and F1-score.