TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Node Classification	Actor	ACM-SGC-1	Accuracy	35.49 ± 1.06	# 34
Node Classification	Actor	ACMII-GCN++	Accuracy	37.09 ± 1.32	# 19
Node Classification	Actor	ACM-GCN++	Accuracy	37.31 ± 1.09	# 17
Node Classification	Actor	ACMII-GCN+	Accuracy	36.14 ± 1.44	# 28
Node Classification	Actor	ACM-GCN+	Accuracy	36.26 ± 1.34	# 25
Node Classification	Actor	ACM-GCN	Accuracy	36.63 ± 0.84	# 22
Node Classification	Actor	ACMII-GCN	Accuracy	36.31 ± 1.2	# 24
Node Classification	Actor	ACM-SGC-2	Accuracy	36.04 ± 0.83	# 31
Node Classification	Chameleon	ACM-SGC-1	Accuracy	63.99 ± 1.66	# 42
Node Classification	Chameleon	ACM-GCN	Accuracy	69.14 ± 1.91	# 29
Node Classification	Chameleon	ACMII-GCN++	Accuracy	74.76 ± 2.2	# 9
Node Classification	Chameleon	ACM-GCN++	Accuracy	74.41 ± 1.49	# 14
Node Classification	Chameleon	ACMII-GCN+	Accuracy	74.56 ± 2.08	# 11
Node Classification	Chameleon	ACM-GCN+	Accuracy	74.47 ± 1.84	# 13
Node Classification	Chameleon	ACM-SGC-2	Accuracy	59.21 ± 2.22	# 49
Node Classification	Chameleon	ACMII-GCN	Accuracy	68.46 ± 1.7	# 33
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	74.41 ± 1.49	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	69.14 ± 1.91	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	68.46 ± 1.7	# 14
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	63.99 ± 1.66	# 20
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	59.21 ± 2.22	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	74.76 ± 2.2	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	74.56 ± 2.08	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	74.47 ± 1.84	# 5
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	75.23 ± 1.72	# 4
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	76.08 ± 2.13	# 1
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	61.66 ± 2.29	# 25
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	58.73 ± 2.52	# 31
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	60.48 ± 1.55	# 28
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	63.68 ± 1.62	# 20
Node Classification	Chameleon (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	64.68 ± 2.85	# 17
Node Classification	Chameleon (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	68.14 ± 1.18	# 10
Node Classification	Chameleon (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	46.72 ± 0.46	# 35
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	75.93 ± 1.71	# 2
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	75.51 ± 1.58	# 3
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	67.53 ± 2.83	# 12
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	67.83 ± 2.63	# 11
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	68.38 ± 1.36	# 8
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	68.4 ± 2.05	# 7
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	68.51 ± 1.7	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	68.4 ± 2.05	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	76.08 ± 2.13	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	MLP-2	1:1 Accuracy	46.72 ± 0.46	# 31
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	58.73 ± 2.52	# 27
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	60.48 ± 1.55	# 25
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	68.51 ± 1.7	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	75.23 ± 1.72	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	61.66 ± 2.29	# 23
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	63.68 ± 1.62	# 19
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	67.53 ± 2.83	# 11
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	67.83 ± 2.63	# 10
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	75.51 ± 1.58	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	64.68 ± 2.85	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	68.14 ± 1.18	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	75.93 ± 1.71	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	68.38 ± 1.36	# 7
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	77.67 ± 1.19	# 2
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	77.2 ± 1.61	# 7
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	77.46 ± 1.65	# 3
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	77.12 ± 1.58	# 12
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	76.73 ± 1.59	# 16
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	76.59 ± 1.69	# 18
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	77.15 ± 1.45	# 8
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	81.58 ± 1.23	# 11
Node Classification	CiteSeer (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	73.77 ± 1.85	# 28
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	82.07 ± 1.04	# 3
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	81.56 ± 1.15	# 13
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	81.32 ± 0.97	# 16
Node Classification	CiteSeer (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	76.25 ± 0.28	# 26
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	80.93 ± 1.16	# 18
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	80.96 ± 0.93	# 17
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	81.83 ± 1.65	# 5
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	81.65 ± 1.48	# 10
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	81.76 ± 1.25	# 8
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	81.87 ± 1.38	# 4
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	81.69 ± 1.25	# 9
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	82.28 ± 1.12	# 2
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	81.79 ± 0.95	# 7
Node Classification	CiteSeer (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	74.49 ± 2.76	# 27
Node Classification	Cora (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	88.01 ± 1.08	# 12
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	88.05 ± 0.99	# 10
Node Classification	Cora (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	88.19 ± 1.17	# 7
Node Classification	Cora (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	88.25 ± 0.96	# 5
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	86.9 ± 1.38	# 20
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	87.69 ± 1.07	# 16
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	88.11 ± 0.96	# 8
Node Classification	Cora (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	87.64 ± 0.99	# 20
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	89.1 ± 1.61	# 9
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	89.00 ± 1.35	# 10
Node Classification	Cora (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	76.44 ± 0.30	# 31
Node Classification	Cora (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	89.52 ± 0.43	# 3
Node Classification	Cora (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	86.63 ± 1.13	# 23
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	89.75 ± 1.16	# 1
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	89.33 ± 0.81	# 6
Node Classification	Cora (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	88.83 ± 1.49	# 16
Node Classification	Cora (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	89.59 ± 1.58	# 2
Node Classification	Cora (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	89.00 ± 0.72	# 10
Node Classification	Cora (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	88.95 ± 1.04	# 13
Node Classification	Cora (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	89.36 ± 1.26	# 5
Node Classification	Cora (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	89.18 ± 1.11	# 8
Node Classification	Cora (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	89.47 ± 1.08	# 4
Node Classification	Cornell	ACMII-GCN	Accuracy	85.95 ± 5.64	# 12
Node Classification	Cornell	ACM-SGC-1	Accuracy	82.43 ± 5.44	# 31
Node Classification	Cornell	ACM-SGC-2	Accuracy	82.43 ± 5.44	# 31
Node Classification	Cornell	ACM-GCN	Accuracy	85.14 ± 6.07	# 20
Node Classification	Cornell	ACM-GCN+	Accuracy	85.68 ± 4.84	# 15
Node Classification	Cornell	ACMII-GCN+	Accuracy	85.41 ± 5.3	# 19
Node Classification	Cornell	ACM-GCN++	Accuracy	85.68 ± 5.8	# 15
Node Classification	Cornell	ACMII-GCN++	Accuracy	86.49 ± 6.73	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	85.68 ± 5.8	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	85.68 ± 4.84	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	86.49 ± 6.73	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	82.43 ± 5.44	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	85.95 ± 5.64	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	82.43 ± 5.44	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	85.14 ± 6.07	# 11
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	85.41 ± 5.3	# 10
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	66.56 ± 13.82	# 31
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	93.93 ± 1.05	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	94.75 ± 3.8	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.08 ± 3.11	# 3
Node Classification	Cornell (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.08 ± 3.11	# 3
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	93.93 ± 1.05	# 7
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 4
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	94.75 ± 3.8	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.26 ± 2.57	# 6
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.44 ± 2.74	# 12
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.62 ± 3.13	# 13
Node Classification	Cornell (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.77 ± 2.17	# 9
Node Classification	Cornell (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.77 ± 1.91	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.9 ± 1.83	# 1
Node Classification	Cornell (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	66.56 ± 13.82	# 33
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	93.93 ± 3.03	# 7
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	93.93 ± 3.03	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	74.43 ± 10.24	# 27
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 2
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	92.62 ± 2.57	# 13
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.62 ± 3.13	# 13
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.44 ± 2.74	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	92.62 ± 2.57	# 13
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	93.61 ± 2.79	# 11
Node Classification	Cornell (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	91.30 ± 0.70	# 18
Node Classification	Cornell (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	74.43 ± 10.24	# 28
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	93.61 ± 2.79	# 11
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.77 ± 2.17	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.77 ± 1.91	# 9
Node Classification	Cornell (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.26 ± 2.57	# 6
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.9 ± 1.83	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACMII-GCN+++	1:1 Accuracy	67.5±0.53	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACMII-GCN+	1:1 Accuracy	67.44±0.31	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCN	1:1 Accuracy	67.01±0.38	# 8
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCNII*	1:1 Accuracy	66.6±0.57	# 13
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCNII	1:1 Accuracy	66.39±0.56	# 17
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-SGC-2	1:1 Accuracy	66.53±0.57	# 15
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-SGC-1	1:1 Accuracy	66.67±0.56	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACMII-GCN	1:1 Accuracy	67.15±0.41	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCN++	1:1 Accuracy	67.3±0.48	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCN+	1:1 Accuracy	67.4±0.44	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	35.49 ± 1.06	# 20
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	36.26 ± 1.34	# 15
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	36.04 ± 0.83	# 18
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	36.14 ± 1.44	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	37.09 ± 1.32	# 10
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	36.31 ± 1.2	# 14
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	36.63 ± 0.84	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	37.31 ± 1.09	# 9
Node Classification	Film (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	41.79 ± 1.01	# 5
Node Classification	Film (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	41.27 ± 1.24	# 12
Node Classification	Film (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	41.37 ± 1.37	# 11
Node Classification	Film (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	40.13 ± 1.21	# 17
Node Classification	Film (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	39.33 ± 1.25	# 18
Node Classification	Film (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	32.72 ± 2.62	# 31
Node Classification	Film (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	35.41 ± 0.97	# 29
Node Classification	Film (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	38.58 ± 0.25	# 22
Node Classification	Film (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	41.86 ± 1.48	# 3
Node Classification	Film (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	41.66 ± 1.42	# 7
Node Classification	Film (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	41.5 ± 1.54	# 9
Node Classification	Film (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	40.31 ± 1.6	# 16
Node Classification	Film (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	41.1 ± 0.75	# 14
Node Classification	Film (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	41.84 ± 1.15	# 4
Node Classification	Film (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	41.27 ± 0.8	# 12
Node Classification	Film (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	41.4 ± 1.23	# 10
Node Classification	genius	ACMII-GCN++	Accuracy	91.01 ± 0.18	# 5
Node Classification	genius	ACM-GCN++	Accuracy	91.37 ± 0.07	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACMII-GCN	1:1 Accuracy	91.19 ± 0.16	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACM-GCN+	1:1 Accuracy	91.33 ± 0.11	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACMII-GCN+	1:1 Accuracy	91.13 ± 0.09	# 6
Node Classification	genius	ACMII-GCN+	Accuracy	91.13 ± 0.09	# 4
Node Classification	genius	ACM-GCN+	Accuracy	91.22 ± 0.13	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACM-GCN++	1:1 Accuracy	91.4 ± 0.07	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACMII-GCN++	1:1 Accuracy	91.01 ± 0.18	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACM-GCN	1:1 Accuracy	91.44 ± 0.08	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACM-GCN++	1:1 Accuracy	86.08 ± 0.43	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACMII-GCN++	1:1 Accuracy	85.95 ± 0.26	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACM-GCN	1:1 Accuracy	82.73 ± 0.52	# 10
Node Classification	Penn94	ACMII-GCN++	Accuracy	85.95 ± 0.26	# 3
Node Classification	Penn94	ACM-GCN++	Accuracy	86.08 ± 0.43	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACMII-GCN	1:1 Accuracy	82.4 ± 0.48	# 12
Node Classification	Penn94	ACMII-GCN+	Accuracy	84.95 ± 0.43	# 7
Node Classification	Penn94	ACM-GCN+	Accuracy	85.05 ± 0.19	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	MLP	1:1 Accuracy	73.61 ± 0.40	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACM-GCN+	1:1 Accuracy	85.05 ± 0.19	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACMII-GCN+	1:1 Accuracy	84.95 ± 0.43	# 6
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	89.01 ± 0.6	# 16
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	89.65 ± 0.58	# 7
Node Classification	PubMed (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	89.71 ± 0.48	# 6
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	88.49 ± 0.51	# 19
Node Classification	PubMed (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	89.78 ± 0.49	# 5
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	89.82 ± 0.41	# 4
Node Classification	PubMed (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	89.89 ± 0.43	# 3
Node Classification	PubMed (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	88.79 ± 0.5	# 24
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	90.66 ± 0.47	# 6
Node Classification	PubMed (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	87.75 ± 0.88	# 27
Node Classification	PubMed (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	89.15 ± 0.87	# 20
Node Classification	PubMed (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	86.43 ± 0.13	# 30
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	90.12 ± 0.4	# 13
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	90.46 ± 0.69	# 10
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	90.18 ± 0.51	# 12
Node Classification	PubMed (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	90.09 ± 0.68	# 14
Node Classification	PubMed (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	91.44 ± 0.59	# 1
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	91.31 ± 0.6	# 2
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	90.96 ± 0.62	# 3
Node Classification	PubMed (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	90.81 ± 0.52	# 4
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	90.74 ± 0.5	# 5
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	90.63 ± 0.56	# 8
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	90.56 ± 0.39	# 9
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	90.39 ± 0.33	# 11
Node Classification	Squirrel	ACM-SGC-1	Accuracy	45.00 ± 1.4	# 41
Node Classification	Squirrel	ACMII-GCN+	Accuracy	67.07 ± 1.65	# 11
Node Classification	Squirrel	ACMII-GCN	Accuracy	51.8 ± 1.5	# 36
Node Classification	Squirrel	ACM-GCN	Accuracy	55.19 ± 1.49	# 31
Node Classification	Squirrel	ACM-GCN+	Accuracy	66.98 ± 1.71	# 13
Node Classification	Squirrel	ACM-GCN++	Accuracy	67.06 ± 1.66	# 12
Node Classification	Squirrel	ACMII-GCN++	Accuracy	67.4 ± 2.21	# 10
Node Classification	Squirrel	ACM-SGC-2	Accuracy	40.02 ± 0.96	# 43
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	55.19 ± 1.49	# 14
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	51.8 ± 1.5	# 18
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	40.02 ± 0.96	# 23
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	45.00 ± 1.4	# 21
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	66.98 ± 1.71	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	67.06 ± 1.66	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	67.4 ± 2.21	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	67.07 ± 1.65	# 4
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	40.9 ± 1.58	# 28
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	69.98 ± 1.53	# 1
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	69.81 ± 1.11	# 2
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	52.31 ± 1.57	# 11
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	53.48 ± 0.6	# 9
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	54.53 ± 2.09	# 8
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	55.73 ± 2.39	# 7
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	55.97 ± 2.03	# 6
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	68.56 ± 1.33	# 4
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	69.26 ± 1.11	# 3
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	38.32 ± 1.5	# 30
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	40.91 ± 1.39	# 27
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	46.4 ± 1.13	# 19
Node Classification	Squirrel (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	52.28 ± 3.61	# 12
Node Classification	Squirrel (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	53.40 ± 1.90	# 10
Node Classification	Squirrel (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	31.28 ± 0.27	# 34
Node Classification	Texas	ACM-SGC-2	Accuracy	81.89 ± 4.53	# 39
Node Classification	Texas	ACM-GCN++	Accuracy	88.38 ± 3.43	# 3
Node Classification	Texas	ACMII-GCN+	Accuracy	88.11 ± 3.24	# 7
Node Classification	Texas	ACMII-GCN	Accuracy	86.76 ± 4.75	# 12
Node Classification	Texas	ACMII-GCN++	Accuracy	88.38 ± 3.43	# 3
Node Classification	Texas	ACM-SGC-1	Accuracy	81.89 ± 4.53	# 39
Node Classification	Texas	ACM-GCN+	Accuracy	88.38 ± 3.64	# 3
Node Classification	Texas	ACM-GCN	Accuracy	87.84 ± 4.4	# 8
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	86.76 ± 4.75	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	88.38 ± 3.43	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	88.38 ± 3.43	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	88.38 ± 3.64	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	81.89 ± 4.53	# 17
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	81.89 ± 4.53	# 17
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	87.84 ± 4.4	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	88.11 ± 3.24	# 4
Node Classification	Texas (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 4
Node Classification	Texas (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.08 ± 2.07	# 5
Node Classification	Texas (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.75 ± 2.41	# 7
Node Classification	Texas (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.74 ± 2.22	# 2
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	96.56 ± 2	# 1
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 6
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.28 ± 2.79	# 13
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.46 ± 1.97	# 16
Node Classification	Texas (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.44 ± 2.54	# 12
Node Classification	Texas (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.61 ± 1.55	# 11
Node Classification	Texas (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	75.41 ± 7.18	# 34
Node Classification	Texas (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	80.66 ± 1.91	# 29
Node Classification	Texas (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	94.75 ± 3.09	# 7
Node Classification	Texas (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	95.41 ± 2.82	# 3
Node Classification	Texas (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	94.75 ± 2.91	# 7
Node Classification	Texas (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	92.26 ± 0.71	# 17
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	80.66 ± 1.91	# 27
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	75.41 ± 7.18	# 31
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	96.56 ± 2	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.74 ± 2.22	# 2
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.44 ± 2.54	# 11
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.75 ± 2.41	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	94.75 ± 3.09	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	94.75 ± 2.91	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.61 ± 1.55	# 10
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.08 ± 2.07	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.28 ± 2.79	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	95.41 ± 2.82	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.46 ± 1.97	# 15
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACMII-GCN	1:1 Accuracy	63.73 ± 0.13	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACM-GCN+	1:1 Accuracy	66.24 ± 0.24	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACMII-GCN+	1:1 Accuracy	65.838 ± 0.153	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACM-GCN++	1:1 Accuracy	65.943 ± 0.284	# 7
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACMII-GCN++	1:1 Accuracy	65.92 ± 0.14	# 8
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACM-GCN	1:1 Accuracy	63.92 ± 0.19	# 14
Node Classification	Wisconsin	ACM-GCN++	Accuracy	88.24 ± 3.16	# 14
Node Classification	Wisconsin	ACM-SGC-2	Accuracy	86.47 ± 3.77	# 30
Node Classification	Wisconsin	ACM-GCN	Accuracy	88.43 ± 3.22	# 9
Node Classification	Wisconsin	ACM-GCN+	Accuracy	88.43 ± 2.39	# 9
Node Classification	Wisconsin	ACMII-GCN+	Accuracy	88.04 ± 3.66	# 15
Node Classification	Wisconsin	ACM-SGC-1	Accuracy	86.47 ± 3.77	# 30
Node Classification	Wisconsin	ACMII-GCN	Accuracy	87.45 ± 3.74	# 22
Node Classification	Wisconsin	ACMII-GCN++	Accuracy	88.43 ± 3.66	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	88.43 ± 3.66	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	86.47 ± 3.77	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	88.04 ± 3.66	# 8
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	87.45 ± 3.74	# 11
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	86.47 ± 3.77	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	88.43 ± 3.22	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	88.43 ± 2.39	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	88.24 ± 3.16	# 7
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	96.62 ± 1.86	# 6
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	96.5 ± 2.08	# 8
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	95.75 ± 2.03	# 10
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	94.00 ± 2.61	# 13
Node Classification	Wisconsin (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	69.50 ± 3.12	# 31
Node Classification	Wisconsin (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	93.87 ± 3.33	# 14
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	96.75 ± 1.79	# 4
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	97.13 ± 1.68	# 2
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	97.5 ± 1.25	# 1
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	97.00 ± 2.63	# 3
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	96.63 ± 2.24	# 5
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	96.62 ± 2.44	# 6
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	96.38 ± 2.59	# 9
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	94.63 ± 2.96	# 11
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	94.37 ± 2.81	# 12
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.25 ± 2.92	# 16
Node Classification	Wisconsin (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	62.50 ± 15.75	# 35
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	94.00 ± 2.61	# 13
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	69.50 ± 3.12	# 28
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	62.50 ± 15.75	# 32
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	97.5 ± 1.25	# 1
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	96.62 ± 2.44	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	97.00 ± 2.63	# 3
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.25 ± 2.92	# 16
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	94.37 ± 2.81	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	94.63 ± 2.96	# 11
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	95.75 ± 2.03	# 10
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	96.38 ± 2.59	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	96.5 ± 2.08	# 8
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	96.62 ± 1.86	# 6
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	96.63 ± 2.24	# 5
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	96.75 ± 1.79	# 4
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	97.13 ± 1.68	# 2

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-chameleon-60-20-20)](https://paperswithcode.com/sota/node-classification-on-chameleon-60-20-20?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-4)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-4?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-cora-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-cora-60-20-20-random?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-7)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-7?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic)](https://paperswithcode.com/sota/node-classification-on-non-homophilic?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-cornell-60-20-20)](https://paperswithcode.com/sota/node-classification-on-cornell-60-20-20?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-6)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-6?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-13)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-13?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-pubmed-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-pubmed-60-20-20-random?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-squirrel-60-20-20)](https://paperswithcode.com/sota/node-classification-on-squirrel-60-20-20?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-9)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-9?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-texas-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-texas-60-20-20-random?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-2)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-2?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-wisconsin-60-20-20)](https://paperswithcode.com/sota/node-classification-on-wisconsin-60-20-20?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-1)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-1?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-citeseer-48-32-20)](https://paperswithcode.com/sota/node-classification-on-citeseer-48-32-20?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-citeseer-60-20-20)](https://paperswithcode.com/sota/node-classification-on-citeseer-60-20-20?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-genius)](https://paperswithcode.com/sota/node-classification-on-genius?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-14)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-14?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-penn94)](https://paperswithcode.com/sota/node-classification-on-penn94?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-11)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-11?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-film-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-film-60-20-20-random?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-pubmed-48-32-20-fixed)](https://paperswithcode.com/sota/node-classification-on-pubmed-48-32-20-fixed?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-12)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-12?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-texas)](https://paperswithcode.com/sota/node-classification-on-texas?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-15)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-15?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-8)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-8?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-cora-48-32-20-fixed)](https://paperswithcode.com/sota/node-classification-on-cora-48-32-20-fixed?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-cornell)](https://paperswithcode.com/sota/node-classification-on-cornell?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-chameleon)](https://paperswithcode.com/sota/node-classification-on-chameleon?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-non-homophilic-10)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-10?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-wisconsin)](https://paperswithcode.com/sota/node-classification-on-wisconsin?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-squirrel)](https://paperswithcode.com/sota/node-classification-on-squirrel?p=revisiting-heterophily-for-graph-neural)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/revisiting-heterophily-for-graph-neural/node-classification-on-actor)](https://paperswithcode.com/sota/node-classification-on-actor?p=revisiting-heterophily-for-graph-neural)`

Revisiting Heterophily For Graph Neural Networks

14 Oct 2022 · Sitao Luan, Chenqing Hua, Qincheng Lu, Jiaqi Zhu, Mingde Zhao, Shuyuan Zhang, Xiao-Wen Chang, Doina Precup ·

Graph Neural Networks (GNNs) extend basic Neural Networks (NNs) by using graph structures based on the relational inductive bias (homophily assumption). While GNNs have been commonly believed to outperform NNs in real-world tasks, recent work has identified a non-trivial set of datasets where their performance compared to NNs is not satisfactory. Heterophily has been considered the main cause of this empirical observation and numerous works have been put forward to address it. In this paper, we first revisit the widely used homophily metrics and point out that their consideration of only graph-label consistency is a shortcoming. Then, we study heterophily from the perspective of post-aggregation node similarity and define new homophily metrics, which are potentially advantageous compared to existing ones. Based on this investigation, we prove that some harmful cases of heterophily can be effectively addressed by local diversification operation. Then, we propose the Adaptive Channel Mixing (ACM), a framework to adaptively exploit aggregation, diversification and identity channels node-wisely to extract richer localized information for diverse node heterophily situations. ACM is more powerful than the commonly used uni-channel framework for node classification tasks on heterophilic graphs and is easy to be implemented in baseline GNN layers. When evaluated on 10 benchmark node classification tasks, ACM-augmented baselines consistently achieve significant performance gain, exceeding state-of-the-art GNNs on most tasks without incurring significant computational burden.

PDF Abstract

Code

Add Remove Mark official

SitaoLuan/ACM-GNN official

Tasks

Add Remove

Inductive Bias

Node Classification

Node Classification on Non-Homophilic (Heterophilic) Graphs

Datasets

Pubmed

Results from the Paper

Edit

Ranked #1 on Node Classification on Cornell (60%/20%/20% random splits)

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Node Classification	Actor	ACM-SGC-1	Accuracy	35.49 ± 1.06	# 34	Compare
Node Classification	Actor	ACMII-GCN++	Accuracy	37.09 ± 1.32	# 19	Compare
Node Classification	Actor	ACM-GCN++	Accuracy	37.31 ± 1.09	# 17	Compare
Node Classification	Actor	ACMII-GCN+	Accuracy	36.14 ± 1.44	# 28	Compare
Node Classification	Actor	ACM-GCN+	Accuracy	36.26 ± 1.34	# 25	Compare
Node Classification	Actor	ACM-GCN	Accuracy	36.63 ± 0.84	# 22	Compare
Node Classification	Actor	ACMII-GCN	Accuracy	36.31 ± 1.2	# 24	Compare
Node Classification	Actor	ACM-SGC-2	Accuracy	36.04 ± 0.83	# 31	Compare
Node Classification	Chameleon	ACM-SGC-1	Accuracy	63.99 ± 1.66	# 42	Compare
Node Classification	Chameleon	ACM-GCN	Accuracy	69.14 ± 1.91	# 29	Compare
Node Classification	Chameleon	ACMII-GCN++	Accuracy	74.76 ± 2.2	# 9	Compare
Node Classification	Chameleon	ACM-GCN++	Accuracy	74.41 ± 1.49	# 14	Compare
Node Classification	Chameleon	ACMII-GCN+	Accuracy	74.56 ± 2.08	# 11	Compare
Node Classification	Chameleon	ACM-GCN+	Accuracy	74.47 ± 1.84	# 13	Compare
Node Classification	Chameleon	ACM-SGC-2	Accuracy	59.21 ± 2.22	# 49	Compare
Node Classification	Chameleon	ACMII-GCN	Accuracy	68.46 ± 1.7	# 33	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	74.41 ± 1.49	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	69.14 ± 1.91	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	68.46 ± 1.7	# 14	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	63.99 ± 1.66	# 20	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	59.21 ± 2.22	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	74.76 ± 2.2	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	74.56 ± 2.08	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	74.47 ± 1.84	# 5	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	75.23 ± 1.72	# 4	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	76.08 ± 2.13	# 1	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	61.66 ± 2.29	# 25	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	58.73 ± 2.52	# 31	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	60.48 ± 1.55	# 28	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	63.68 ± 1.62	# 20	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	64.68 ± 2.85	# 17	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	68.14 ± 1.18	# 10	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	46.72 ± 0.46	# 35	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	75.93 ± 1.71	# 2	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	75.51 ± 1.58	# 3	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	67.53 ± 2.83	# 12	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	67.83 ± 2.63	# 11	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	68.38 ± 1.36	# 8	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	68.4 ± 2.05	# 7	Compare
Node Classification	Chameleon (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	68.51 ± 1.7	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	68.4 ± 2.05	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	76.08 ± 2.13	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	MLP-2	1:1 Accuracy	46.72 ± 0.46	# 31	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	58.73 ± 2.52	# 27	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	60.48 ± 1.55	# 25	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	68.51 ± 1.7	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	75.23 ± 1.72	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	61.66 ± 2.29	# 23	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	63.68 ± 1.62	# 19	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	67.53 ± 2.83	# 11	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	67.83 ± 2.63	# 10	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	75.51 ± 1.58	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	64.68 ± 2.85	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	68.14 ± 1.18	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	75.93 ± 1.71	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	68.38 ± 1.36	# 7	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	77.67 ± 1.19	# 2	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	77.2 ± 1.61	# 7	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	77.46 ± 1.65	# 3	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	77.12 ± 1.58	# 12	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	76.73 ± 1.59	# 16	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	76.59 ± 1.69	# 18	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	77.15 ± 1.45	# 8	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	81.58 ± 1.23	# 11	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	73.77 ± 1.85	# 28	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	82.07 ± 1.04	# 3	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	81.56 ± 1.15	# 13	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	81.32 ± 0.97	# 16	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	76.25 ± 0.28	# 26	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	80.93 ± 1.16	# 18	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	80.96 ± 0.93	# 17	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	81.83 ± 1.65	# 5	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	81.65 ± 1.48	# 10	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	81.76 ± 1.25	# 8	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	81.87 ± 1.38	# 4	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	81.69 ± 1.25	# 9	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	82.28 ± 1.12	# 2	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	81.79 ± 0.95	# 7	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	74.49 ± 2.76	# 27	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	88.01 ± 1.08	# 12	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	88.05 ± 0.99	# 10	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	88.19 ± 1.17	# 7	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	88.25 ± 0.96	# 5	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	86.9 ± 1.38	# 20	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	87.69 ± 1.07	# 16	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	88.11 ± 0.96	# 8	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	87.64 ± 0.99	# 20	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	89.1 ± 1.61	# 9	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	89.00 ± 1.35	# 10	Compare
Node Classification	Cora (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	76.44 ± 0.30	# 31	Compare
Node Classification	Cora (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	89.52 ± 0.43	# 3	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	86.63 ± 1.13	# 23	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	89.75 ± 1.16	# 1	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	89.33 ± 0.81	# 6	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	88.83 ± 1.49	# 16	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	89.59 ± 1.58	# 2	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	89.00 ± 0.72	# 10	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	88.95 ± 1.04	# 13	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	89.36 ± 1.26	# 5	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	89.18 ± 1.11	# 8	Compare
Node Classification	Cora (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	89.47 ± 1.08	# 4	Compare
Node Classification	Cornell	ACMII-GCN	Accuracy	85.95 ± 5.64	# 12	Compare
Node Classification	Cornell	ACM-SGC-1	Accuracy	82.43 ± 5.44	# 31	Compare
Node Classification	Cornell	ACM-SGC-2	Accuracy	82.43 ± 5.44	# 31	Compare
Node Classification	Cornell	ACM-GCN	Accuracy	85.14 ± 6.07	# 20	Compare
Node Classification	Cornell	ACM-GCN+	Accuracy	85.68 ± 4.84	# 15	Compare
Node Classification	Cornell	ACMII-GCN+	Accuracy	85.41 ± 5.3	# 19	Compare
Node Classification	Cornell	ACM-GCN++	Accuracy	85.68 ± 5.8	# 15	Compare
Node Classification	Cornell	ACMII-GCN++	Accuracy	86.49 ± 6.73	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	85.68 ± 5.8	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	85.68 ± 4.84	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	86.49 ± 6.73	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	82.43 ± 5.44	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	85.95 ± 5.64	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	82.43 ± 5.44	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	85.14 ± 6.07	# 11	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	85.41 ± 5.3	# 10	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	66.56 ± 13.82	# 31	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	93.93 ± 1.05	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	94.75 ± 3.8	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.08 ± 3.11	# 3	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.08 ± 3.11	# 3	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	93.93 ± 1.05	# 7	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 4	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	94.75 ± 3.8	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.26 ± 2.57	# 6	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.44 ± 2.74	# 12	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.62 ± 3.13	# 13	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.77 ± 2.17	# 9	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.77 ± 1.91	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.9 ± 1.83	# 1	Compare
Node Classification	Cornell (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	66.56 ± 13.82	# 33	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	93.93 ± 3.03	# 7	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	93.93 ± 3.03	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	74.43 ± 10.24	# 27	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 2	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	92.62 ± 2.57	# 13	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.62 ± 3.13	# 13	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.44 ± 2.74	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	92.62 ± 2.57	# 13	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	93.61 ± 2.79	# 11	Compare
Node Classification	Cornell (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	91.30 ± 0.70	# 18	Compare
Node Classification	Cornell (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	74.43 ± 10.24	# 28	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	93.61 ± 2.79	# 11	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.77 ± 2.17	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.77 ± 1.91	# 9	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.26 ± 2.57	# 6	Compare
Node Classification	Cornell (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.9 ± 1.83	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACMII-GCN+++	1:1 Accuracy	67.5±0.53	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACMII-GCN+	1:1 Accuracy	67.44±0.31	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCN	1:1 Accuracy	67.01±0.38	# 8	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCNII*	1:1 Accuracy	66.6±0.57	# 13	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCNII	1:1 Accuracy	66.39±0.56	# 17	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-SGC-2	1:1 Accuracy	66.53±0.57	# 15	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-SGC-1	1:1 Accuracy	66.67±0.56	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACMII-GCN	1:1 Accuracy	67.15±0.41	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCN++	1:1 Accuracy	67.3±0.48	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	ACM-GCN+	1:1 Accuracy	67.4±0.44	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	35.49 ± 1.06	# 20	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	36.26 ± 1.34	# 15	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	36.04 ± 0.83	# 18	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	36.14 ± 1.44	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	37.09 ± 1.32	# 10	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	36.31 ± 1.2	# 14	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	36.63 ± 0.84	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	37.31 ± 1.09	# 9	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	41.79 ± 1.01	# 5	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	41.27 ± 1.24	# 12	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	41.37 ± 1.37	# 11	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	40.13 ± 1.21	# 17	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	39.33 ± 1.25	# 18	Compare
Node Classification	Film (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	32.72 ± 2.62	# 31	Compare
Node Classification	Film (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	35.41 ± 0.97	# 29	Compare
Node Classification	Film (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	38.58 ± 0.25	# 22	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	41.86 ± 1.48	# 3	Compare
Node Classification	Film (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	41.66 ± 1.42	# 7	Compare
Node Classification	Film (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	41.5 ± 1.54	# 9	Compare
Node Classification	Film (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	40.31 ± 1.6	# 16	Compare
Node Classification	Film (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	41.1 ± 0.75	# 14	Compare
Node Classification	Film (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	41.84 ± 1.15	# 4	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	41.27 ± 0.8	# 12	Compare
Node Classification	Film (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	41.4 ± 1.23	# 10	Compare
Node Classification	genius	ACMII-GCN++	Accuracy	91.01 ± 0.18	# 5	Compare
Node Classification	genius	ACM-GCN++	Accuracy	91.37 ± 0.07	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACMII-GCN	1:1 Accuracy	91.19 ± 0.16	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACM-GCN+	1:1 Accuracy	91.33 ± 0.11	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACMII-GCN+	1:1 Accuracy	91.13 ± 0.09	# 6	Compare
Node Classification	genius	ACMII-GCN+	Accuracy	91.13 ± 0.09	# 4	Compare
Node Classification	genius	ACM-GCN+	Accuracy	91.22 ± 0.13	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACM-GCN++	1:1 Accuracy	91.4 ± 0.07	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACMII-GCN++	1:1 Accuracy	91.01 ± 0.18	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	ACM-GCN	1:1 Accuracy	91.44 ± 0.08	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACM-GCN++	1:1 Accuracy	86.08 ± 0.43	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACMII-GCN++	1:1 Accuracy	85.95 ± 0.26	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACM-GCN	1:1 Accuracy	82.73 ± 0.52	# 10	Compare
Node Classification	Penn94	ACMII-GCN++	Accuracy	85.95 ± 0.26	# 3	Compare
Node Classification	Penn94	ACM-GCN++	Accuracy	86.08 ± 0.43	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACMII-GCN	1:1 Accuracy	82.4 ± 0.48	# 12	Compare
Node Classification	Penn94	ACMII-GCN+	Accuracy	84.95 ± 0.43	# 7	Compare
Node Classification	Penn94	ACM-GCN+	Accuracy	85.05 ± 0.19	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	MLP	1:1 Accuracy	73.61 ± 0.40	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACM-GCN+	1:1 Accuracy	85.05 ± 0.19	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	ACMII-GCN+	1:1 Accuracy	84.95 ± 0.43	# 6	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	89.01 ± 0.6	# 16	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	89.65 ± 0.58	# 7	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	89.71 ± 0.48	# 6	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	88.49 ± 0.51	# 19	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	89.78 ± 0.49	# 5	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	89.82 ± 0.41	# 4	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	89.89 ± 0.43	# 3	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	88.79 ± 0.5	# 24	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	90.66 ± 0.47	# 6	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	87.75 ± 0.88	# 27	Compare
Node Classification	PubMed (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	89.15 ± 0.87	# 20	Compare
Node Classification	PubMed (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	86.43 ± 0.13	# 30	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	90.12 ± 0.4	# 13	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	90.46 ± 0.69	# 10	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	90.18 ± 0.51	# 12	Compare
Node Classification	PubMed (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	90.09 ± 0.68	# 14	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	91.44 ± 0.59	# 1	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	91.31 ± 0.6	# 2	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	90.96 ± 0.62	# 3	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	90.81 ± 0.52	# 4	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	90.74 ± 0.5	# 5	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	90.63 ± 0.56	# 8	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	90.56 ± 0.39	# 9	Compare
Node Classification	PubMed (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	90.39 ± 0.33	# 11	Compare
Node Classification	Squirrel	ACM-SGC-1	Accuracy	45.00 ± 1.4	# 41	Compare
Node Classification	Squirrel	ACMII-GCN+	Accuracy	67.07 ± 1.65	# 11	Compare
Node Classification	Squirrel	ACMII-GCN	Accuracy	51.8 ± 1.5	# 36	Compare
Node Classification	Squirrel	ACM-GCN	Accuracy	55.19 ± 1.49	# 31	Compare
Node Classification	Squirrel	ACM-GCN+	Accuracy	66.98 ± 1.71	# 13	Compare
Node Classification	Squirrel	ACM-GCN++	Accuracy	67.06 ± 1.66	# 12	Compare
Node Classification	Squirrel	ACMII-GCN++	Accuracy	67.4 ± 2.21	# 10	Compare
Node Classification	Squirrel	ACM-SGC-2	Accuracy	40.02 ± 0.96	# 43	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	55.19 ± 1.49	# 14	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	51.8 ± 1.5	# 18	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	40.02 ± 0.96	# 23	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	45.00 ± 1.4	# 21	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	66.98 ± 1.71	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	67.06 ± 1.66	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	67.4 ± 2.21	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	67.07 ± 1.65	# 4	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	40.9 ± 1.58	# 28	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	69.98 ± 1.53	# 1	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	69.81 ± 1.11	# 2	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	52.31 ± 1.57	# 11	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	53.48 ± 0.6	# 9	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	54.53 ± 2.09	# 8	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	55.73 ± 2.39	# 7	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	55.97 ± 2.03	# 6	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	68.56 ± 1.33	# 4	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	69.26 ± 1.11	# 3	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	38.32 ± 1.5	# 30	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	40.91 ± 1.39	# 27	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	46.4 ± 1.13	# 19	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	52.28 ± 3.61	# 12	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	53.40 ± 1.90	# 10	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	31.28 ± 0.27	# 34	Compare
Node Classification	Texas	ACM-SGC-2	Accuracy	81.89 ± 4.53	# 39	Compare
Node Classification	Texas	ACM-GCN++	Accuracy	88.38 ± 3.43	# 3	Compare
Node Classification	Texas	ACMII-GCN+	Accuracy	88.11 ± 3.24	# 7	Compare
Node Classification	Texas	ACMII-GCN	Accuracy	86.76 ± 4.75	# 12	Compare
Node Classification	Texas	ACMII-GCN++	Accuracy	88.38 ± 3.43	# 3	Compare
Node Classification	Texas	ACM-SGC-1	Accuracy	81.89 ± 4.53	# 39	Compare
Node Classification	Texas	ACM-GCN+	Accuracy	88.38 ± 3.64	# 3	Compare
Node Classification	Texas	ACM-GCN	Accuracy	87.84 ± 4.4	# 8	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	86.76 ± 4.75	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	88.38 ± 3.43	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	88.38 ± 3.43	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	88.38 ± 3.64	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	81.89 ± 4.53	# 17	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	81.89 ± 4.53	# 17	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	87.84 ± 4.4	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	88.11 ± 3.24	# 4	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 4	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.08 ± 2.07	# 5	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.75 ± 2.41	# 7	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.74 ± 2.22	# 2	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	96.56 ± 2	# 1	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 6	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.28 ± 2.79	# 13	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.46 ± 1.97	# 16	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.44 ± 2.54	# 12	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.61 ± 1.55	# 11	Compare
Node Classification	Texas (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	75.41 ± 7.18	# 34	Compare
Node Classification	Texas (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	80.66 ± 1.91	# 29	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	94.75 ± 3.09	# 7	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	95.41 ± 2.82	# 3	Compare
Node Classification	Texas (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	94.75 ± 2.91	# 7	Compare
Node Classification	Texas (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	92.26 ± 0.71	# 17	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	94.92 ± 2.79	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	80.66 ± 1.91	# 27	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	75.41 ± 7.18	# 31	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	96.56 ± 2	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	95.74 ± 2.22	# 2	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	93.44 ± 2.54	# 11	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	94.75 ± 2.41	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	94.75 ± 3.09	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	94.75 ± 2.91	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.61 ± 1.55	# 10	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	95.08 ± 2.07	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	95.25 ± 1.55	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	93.28 ± 2.79	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	95.41 ± 2.82	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	92.46 ± 1.97	# 15	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACMII-GCN	1:1 Accuracy	63.73 ± 0.13	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACM-GCN+	1:1 Accuracy	66.24 ± 0.24	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACMII-GCN+	1:1 Accuracy	65.838 ± 0.153	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACM-GCN++	1:1 Accuracy	65.943 ± 0.284	# 7	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACMII-GCN++	1:1 Accuracy	65.92 ± 0.14	# 8	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	ACM-GCN	1:1 Accuracy	63.92 ± 0.19	# 14	Compare
Node Classification	Wisconsin	ACM-GCN++	Accuracy	88.24 ± 3.16	# 14	Compare
Node Classification	Wisconsin	ACM-SGC-2	Accuracy	86.47 ± 3.77	# 30	Compare
Node Classification	Wisconsin	ACM-GCN	Accuracy	88.43 ± 3.22	# 9	Compare
Node Classification	Wisconsin	ACM-GCN+	Accuracy	88.43 ± 2.39	# 9	Compare
Node Classification	Wisconsin	ACMII-GCN+	Accuracy	88.04 ± 3.66	# 15	Compare
Node Classification	Wisconsin	ACM-SGC-1	Accuracy	86.47 ± 3.77	# 30	Compare
Node Classification	Wisconsin	ACMII-GCN	Accuracy	87.45 ± 3.74	# 22	Compare
Node Classification	Wisconsin	ACMII-GCN++	Accuracy	88.43 ± 3.66	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACMII-GCN++	1:1 Accuracy	88.43 ± 3.66	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-SGC-1	1:1 Accuracy	86.47 ± 3.77	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACMII-GCN+	1:1 Accuracy	88.04 ± 3.66	# 8	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACMII-GCN	1:1 Accuracy	87.45 ± 3.74	# 11	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-SGC-2	1:1 Accuracy	86.47 ± 3.77	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-GCN	1:1 Accuracy	88.43 ± 3.22	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-GCN+	1:1 Accuracy	88.43 ± 2.39	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	ACM-GCN++	1:1 Accuracy	88.24 ± 3.16	# 7	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	96.62 ± 1.86	# 6	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	96.5 ± 2.08	# 8	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	95.75 ± 2.03	# 10	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	94.00 ± 2.61	# 13	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	69.50 ± 3.12	# 31	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	MLP-2	1:1 Accuracy	93.87 ± 3.33	# 14	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	96.75 ± 1.79	# 4	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	97.13 ± 1.68	# 2	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	97.5 ± 1.25	# 1	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	97.00 ± 2.63	# 3	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	96.63 ± 2.24	# 5	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	96.62 ± 2.44	# 6	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	96.38 ± 2.59	# 9	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	94.63 ± 2.96	# 11	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	94.37 ± 2.81	# 12	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.25 ± 2.92	# 16	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	62.50 ± 15.75	# 35	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-SGC-2	1:1 Accuracy	94.00 ± 2.61	# 13	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	GAT+JK	1:1 Accuracy	69.50 ± 3.12	# 28	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	GCN+JK	1:1 Accuracy	62.50 ± 15.75	# 32	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCN++	1:1 Accuracy	97.5 ± 1.25	# 1	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-GCN	1:1 Accuracy	96.62 ± 2.44	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-Snowball-3	1:1 Accuracy	97.00 ± 2.63	# 3	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-SGC-1	1:1 Accuracy	93.25 ± 2.92	# 16	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCNII*	1:1 Accuracy	94.37 ± 2.81	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCNII	1:1 Accuracy	94.63 ± 2.96	# 11	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCN	1:1 Accuracy	95.75 ± 2.03	# 10	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-Snowball-2	1:1 Accuracy	96.38 ± 2.59	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-GCN+	1:1 Accuracy	96.5 ± 2.08	# 8	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACM-Snowball-3	1:1 Accuracy	96.62 ± 1.86	# 6	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-Snowball-2	1:1 Accuracy	96.63 ± 2.24	# 5	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-GCN+	1:1 Accuracy	96.75 ± 1.79	# 4	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	ACMII-GCN++	1:1 Accuracy	97.13 ± 1.68	# 2	Compare

Methods

Add Remove

No methods listed for this paper. Add relevant methods here

Edit Social Preview

Revisiting Heterophily For Graph Neural Networks

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit

Methods Edit Add Remove

Code

Add Remove Mark official

Tasks

Add Remove

Datasets

Results from the Paper

Edit

Methods

Add Remove