Table 2 Model architectures and... | Emerald Publishing

Table 2

Model architectures and hyperparameters

Model	Layers	Key hyperparameters	Regularization	Optimizer/LR	Epochs	Early stop
GRU-RNN	[GRU64 → GRU32] → Dense16 → Sigmoid	recurrent_dropout = 0.2	L2 = 1e−5 Dropout on dense = 0.2	Adam 1e−3	50	yes (AUC-PR)
1D-CNN	2×[Conv1D {32,64}, k = {3,5} → BN → ReLU → MaxPool(2)] → Dense64 → Dropout0.3 → Sigmoid	stride = 1; padding = ”same”	Dropout = 0.3	Adam 1e−3	50	yes (AUC-PR)
MLP (baseline)	128 → 64→32, ReLU	weighted BCE	Dropout = 0.3

Model	Layers	Key hyperparameters	Regularization	Optimizer/LR	Epochs	Early stop
GRU-RNN	[GRU64 → GRU32] → Dense16 → Sigmoid	recurrent_dropout = 0.2	L2 = 1e−5Dropout on dense = 0.2	Adam 1e−3	50	yes (AUC-PR)
1D-CNN	2×[Conv1D {32,64}, k = {3,5} → BN → ReLU → MaxPool(2)] → Dense64 → Dropout0.3 → Sigmoid	stride = 1; padding = ”same”	Dropout = 0.3	Adam 1e−3	50	yes (AUC-PR)
MLP (baseline)	128 → 64→32, ReLU	weighted BCE	Dropout = 0.3