Table 3:

AUC (%) and AP (%) scores for k-GAEs with k = 1, 2, 6, 36, 72, and ${\text{DGAE}}_{\alpha }^{\beta }$ with layers 6, 36, and 72. Here, the p-value represents the statistical significance of the difference between the best performance of our method (reported in the last 3 rows) and the best performance of other methods.

Dataset	chameleon		texas		wisconsin
	AUC (%)	AP (%)	AUC (%)	AP (%)	AUC (%)	AP (%)
Linear-GAE	45.21 ± 3.75	53.74 ± 3.79	12.97 ± 11.78	40.41 ± 8.45	21.98 ± 7.24	44.53 ± 5.64
2-GAE	46.33 ± 6.26	54.62 ± 5.77	22.19 ± 12.65	46.35 ± 8.55	34.63 ± 11.13	57.09 ± 7.12
6-GAE	37.67 ± 5.80	49.71 ± 4.69	15.31 ± 10.66	42.38 ± 8.73	19.51 ± 7.88	45.14 ± 8.00
36-GAE	50.00 ± 0.00	50.00 ± 0.00	50.00±0.00	50.00 ± 0.00	50.00 ± 0.00	50.00 ± 0.00
72-GAE	50.00 ± 0.00	50.00 ± 0.00	50.00±0.00	50.00 ± 0.00	50.00 ± 0.00	50.00 ± 0.00
$6{\text{-DGAE}}_{\alpha }^{\beta }$	56.89 ± 4.14	62.89 ± 4.42	45.78 ± 9.63	61.06±8.74	53.95 ± 10.92	67.76 ± 8.62
$36{\text{-DGAE}}_{\alpha }^{\beta }$	58.35 ± 6.61	60.60 ± 5.78	45.16 ± 10.86	60.44 ± 9.74	54.20±10.55	67.78±8.73
$72{\text{-DGAE}}_{\alpha }^{\beta }$	62.26±0.17	65.02±1.46	46.09 ± 9.56	61.04 ± 8.82	53.70 ± 9.52	67.46 ± 7.80
(t-statistics, p-value)	(228.06, <0.0001)	(5.52, <0.0001)	(−1.29, 0.2122)	(4.00, 0.0008)	(1.26, 0.2241)	(3.00, 0.0077)