Table 2.

Essential hyperparameters used in the proposed method.

Parameter	Value
Learning rate (γ)	0.01
Latent regulator size	8×8
Drop rate	0.2
Number of layers	8
Units in each layer	16
Weight decay (λ)	0.005