Fig 1.
The proposed model, which is divided into three parts.
The dual Graph convolution network (Dual-GCN) module, Clustering-Optimized module, and Positive/Negative Features Enhancement Network (P/N-FEN) module. A and B represent adjacency matrices, respectively.
Table 1.
Overview of datasets.
Table 2.
Performance comparison with proposed methods on dierent datasets.
Table 3.
Ablation study of loss function.
Fig 2.
The influence of the value lr on AUC on three datasets.
Fig 3.
Impact of hyperparameter on model performance.
(a) shows the impact of on model’s AUC performance, (b) shows the impact of
on model’s RMSE performance.
Fig 4.
The impact of label noise on model performance.