Table 1.

The settings of hyperparameters for our proposed model AMMVF-DTI.

Hyperparameter	Value
Epoch	40
Dropout	0.1
Learning rate	1 × 10−3
Regularization coefficient	1 × 10−4
The radius	2
The n-gram	3
The number of major potential associations K	16
The dimensions of the hidden layer	64
The number of GAT layers	3
The number of multi-head self-attention	8