Table 5.

Accuracy of semi-supervised node classification on Pubmed.

Method	90%	80%	70%	60%	50%	40%	30%	20%	10%
GCN	0.871	0.838	0.838	0.806	0.806	0.774	0.774	0.741	0.741
αLoNGAE	0.807	0.803	0.800	0.797	0.796	0.793	0.790	0.787	0.786
GAT	0.794	0.792	0.790	0.788	0.786	0.784	0.782	0.780	0.788
MT-GAT (ours)	0.854	0.847	0.843	0.836	0.831	0.824	0.822	0.816	0.806

The best results are shown in bold, and our MT-GAT with significant improvements over the baselines is shown with underlines.