Node Classification on Non-Homophilic (Heterophilic) Graphs 벤치마크

20% random splits)

33개 결과 · ⬇ CSV · JSON

1:1 Accuracy

Rank	Model	1:1 Accuracy	Paper	Code	Year
1	ACMII-GCN	95.9 ± 1.83	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
2	ACMII-Snowball-2	95.25 ± 1.55	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
3	ACM-Snowball-2	95.08 ± 3.11	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
4	ACM-GCN+	94.92 ± 2.79	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
5	ACM-GCN	94.75 ± 3.8	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
6	ACM-Snowball-3	94.26 ± 2.57	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
7	ACM-GCN++	93.93 ± 1.05	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
8	ACMII-GCN+	93.93 ± 3.03	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
9	ACM-SGC-2	93.77 ± 2.17	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
10	ACM-SGC-1	93.77 ± 1.91	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
11	ACMII-Snowball-3	93.61 ± 2.79	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
12	ACM-GCNII*	93.44 ± 2.74	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
13	ACMII-GCN++	92.62 ± 2.57	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
14	ACM-GCNII	92.62 ± 3.13	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
15	BernNet	92.13 ± 1.64	BernNet: Learning Arbitrary Graph Spectral Filters via Bernstein Approximation	ivam-he/BernNet	2021
16	APPNP	91.80 ± 0.63	Predict then Propagate: Graph Neural Networks meet Personalized PageRank	dmlc/dgl · dmlc/dgl · benedekrozemberczki/APPNP · +2	2018
17	GPRGNN	91.36 ± 0.70	Adaptive Universal Generalized PageRank Graph Neural Network	jianhao2016/GPRGNN	2020
18	MLP-2	91.30 ± 0.70	Adaptive Universal Generalized PageRank Graph Neural Network	jianhao2016/GPRGNN	2020
19	GCNII*	90.49 ± 4.45	Simple and Deep Graph Convolutional Networks	chennnM/GCNII · chennnM/GCNII · zhanglab-aim/cancer-net · +1	2020
20	GCNII	89.18 ± 3.96	Simple and Deep Graph Convolutional Networks	chennnM/GCNII · chennnM/GCNII · zhanglab-aim/cancer-net · +1	2020
21	FAGCN	88.03 ± 5.6	Beyond Low-frequency Information in Graph Convolutional Networks	bdy9527/FAGCN	2021
22	H2GCN	86.23 ± 4.71	Beyond Homophily in Graph Neural Networks: Current Limitations and Effective Designs	GemsLab/H2GCN · GitEventhandler/H2GCN-PyTorch · sxwee/GNNsIMPL · +1	2020
23	Snowball-3	82.95 ± 2.1	Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks	PwnerHarry/Stronger_GCN	2019
24	Snowball-2	82.62 ± 2.34	Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks	PwnerHarry/Stronger_GCN	2019
25	GCN	82.46 ± 3.11	Semi-Supervised Classification with Graph Convolutional Networks	dmlc/dgl · dmlc/dgl · tkipf/gcn · +52	2016
26	GAT	76.00 ± 1.01	Graph Attention Networks	labmlai/annotated_deep_learning_paper_implementations · dmlc/dgl · dmlc/dgl · +90	2017
27	GAT+JK	74.43 ± 10.24	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
28	SGC-2	72.62 ± 9.92	Simplifying Graph Convolutional Networks	dmlc/dgl · dmlc/dgl · dmlc/dgl · +4	2019
29	GraphSAGE	71.41 ± 1.24	Inductive Representation Learning on Large Graphs	pyg-team/pytorch_geometric · dmlc/dgl · dmlc/dgl · +17	2017
30	SGC-1	70.98 ± 8.39	Simplifying Graph Convolutional Networks	dmlc/dgl · dmlc/dgl · dmlc/dgl · +4	2019
31	GCN+JK	66.56 ± 13.82	Revisiting Heterophily For Graph Neural Networks	SitaoLuan/ACM-GNN	2022
32	Geom-GCN*	60.81	Geom-GCN: Geometric Graph Convolutional Networks	bingzhewei/geom-gcn · graphdml-uiuc-jlu/geom-gcn · alexfanjn/geomgcn_pyg · +1	2020
33	MixHop	60.33 ± 28.53	MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing	dmlc/dgl · benedekrozemberczki/MixHop-and-N-GCN · samihaija/mixhop	2019