TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Node Classification	Actor	NLMLP	Accuracy	37.9 ± 1.3	# 8
Node Classification	Actor	NLGCN	Accuracy	31.6 ± 1.0	# 45
Node Classification	Actor	NLGAT	Accuracy	29.5 ± 1.3	# 47
Node Classification	Chameleon	NLGAT	Accuracy	65.7 ± 1.4	# 38
Node Classification	Chameleon	NLMLP	Accuracy	50.7 ± 2.2	# 52
Node Classification	Chameleon	NLGCN	Accuracy	70.1 ± 2.9	# 24
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	50.7 ± 2.2	# 27
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	70.1 ± 2.9	# 10
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	65.7 ± 1.4	# 18
Node Classification	Citeseer (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	76.2 ± 1.6	# 21
Node Classification	Citeseer (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	73.4 ± 1.9	# 24
Node Classification	Citeseer (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	75.2 ± 1.4	# 22
Node Classification	Cora (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	88.5 ± 1.8	# 1
Node Classification	Cora (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	88.1 ± 1.0	# 9
Node Classification	Cora (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	76.9 ± 1.8	# 25
Node Classification	Cornell	NLGCN	Accuracy	57.6 ± 5.5	# 48
Node Classification	Cornell	NLMLP	Accuracy	84.9 ± 5.7	# 21
Node Classification	Cornell	NLGAT	Accuracy	54.7 ± 7.6	# 51
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	84.9 ± 5.7	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	54.7 ± 7.6	# 27
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	57.6 ± 5.5	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	31.6 ± 1.0	# 24
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	29.5 ± 1.3	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	37.9 ± 1.3	# 1
Node Classification	PubMed (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	88.2 ± 0.5	# 20
Node Classification	PubMed (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	88.2 ± 0.3	# 20
Node Classification	PubMed (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	89.0 ± 0.5	# 17
Node Classification	Squirrel	NLMLP	Accuracy	33.7 ± 1.5	# 48
Node Classification	Squirrel	NLGCN	Accuracy	59.0 ± 1.2	# 23
Node Classification	Squirrel	NLGAT	Accuracy	56.8 ± 2.5	# 27
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	33.7 ± 1.5	# 27
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	56.8 ± 2.5	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	59.0 ± 1.2	# 9
Node Classification	Texas	NLGAT	Accuracy	62.6 ± 7.1	# 51
Node Classification	Texas	NLGCN	Accuracy	65.5 ± 6.6	# 49
Node Classification	Texas	NLMLP	Accuracy	85.4 ± 3.8	# 24
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	62.6 ± 7.1	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	85.4 ± 3.8	# 9
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	65.5 ± 6.6	# 25
Node Classification	Wisconsin	NLMLP	Accuracy	87.3 ± 4.3	# 23
Node Classification	Wisconsin	NLGCN	Accuracy	60.2 ± 5.3	# 50
Node Classification	Wisconsin	NLGAT	Accuracy	56.9 ± 7.3	# 52
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	56.9 ± 7.3	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	60.2 ± 5.3	# 25
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	87.3 ± 4.3	# 12

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-cora-48-32-20-fixed)](https://paperswithcode.com/sota/node-classification-on-cora-48-32-20-fixed?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-non-homophilic-10)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-10?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-actor)](https://paperswithcode.com/sota/node-classification-on-actor?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-non-homophilic-12)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-12?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-non-homophilic-9)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-9?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-non-homophilic-11)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-11?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-non-homophilic-7)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-7?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-non-homophilic-8)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-8?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-pubmed-48-32-20-fixed)](https://paperswithcode.com/sota/node-classification-on-pubmed-48-32-20-fixed?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-citeseer-48-32-20)](https://paperswithcode.com/sota/node-classification-on-citeseer-48-32-20?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-cornell)](https://paperswithcode.com/sota/node-classification-on-cornell?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-squirrel)](https://paperswithcode.com/sota/node-classification-on-squirrel?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-wisconsin)](https://paperswithcode.com/sota/node-classification-on-wisconsin?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-chameleon)](https://paperswithcode.com/sota/node-classification-on-chameleon?p=non-local-graph-neural-networks)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/non-local-graph-neural-networks/node-classification-on-texas)](https://paperswithcode.com/sota/node-classification-on-texas?p=non-local-graph-neural-networks)`

Non-Local Graph Neural Networks

29 May 2020 · Meng Liu, Zhengyang Wang, Shuiwang Ji ·

Modern graph neural networks (GNNs) learn node embeddings through multilayer local aggregation and achieve great success in applications on assortative graphs. However, tasks on disassortative graphs usually require non-local aggregation. In addition, we find that local aggregation is even harmful for some disassortative graphs. In this work, we propose a simple yet effective non-local aggregation framework with an efficient attention-guided sorting for GNNs. Based on it, we develop various non-local GNNs. We perform thorough experiments to analyze disassortative graph datasets and evaluate our non-local GNNs. Experimental results demonstrate that our non-local GNNs significantly outperform previous state-of-the-art methods on seven benchmark datasets of disassortative graphs, in terms of both model performance and efficiency.

PDF Abstract

Code

Add Remove Mark official

divelab/Non-Local-GNN official

Tasks

Add Remove

Node Classification

Node Classification on Non-Homophilic (Heterophilic) Graphs

Datasets

Pubmed

OGB

Wiki Squirrel WebKB Cornell Chameleon (48%/32%/20% fixed splits) Squirrel (48%/32%/20% fixed splits) Cornell (48%/32%/20% fixed splits) Wisconsin (48%/32%/20% fixed splits) Citeseer (48%/32%/20% fixed splits) PubMed (48%/32%/20% fixed splits) Cora (48%/32%/20% fixed splits) Film(48%/32%/20% fixed splits) Texas (48%/32%/20% fixed splits)

Results from the Paper

Edit

Ranked #1 on Node Classification on Cora (48%/32%/20% fixed splits)

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Node Classification	Actor	NLMLP	Accuracy	37.9 ± 1.3	# 8	Compare
Node Classification	Actor	NLGCN	Accuracy	31.6 ± 1.0	# 45	Compare
Node Classification	Actor	NLGAT	Accuracy	29.5 ± 1.3	# 47	Compare
Node Classification	Chameleon	NLGAT	Accuracy	65.7 ± 1.4	# 38	Compare
Node Classification	Chameleon	NLMLP	Accuracy	50.7 ± 2.2	# 52	Compare
Node Classification	Chameleon	NLGCN	Accuracy	70.1 ± 2.9	# 24	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	50.7 ± 2.2	# 27	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	70.1 ± 2.9	# 10	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	65.7 ± 1.4	# 18	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	76.2 ± 1.6	# 21	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	73.4 ± 1.9	# 24	Compare
Node Classification	Citeseer (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	75.2 ± 1.4	# 22	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	88.5 ± 1.8	# 1	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	88.1 ± 1.0	# 9	Compare
Node Classification	Cora (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	76.9 ± 1.8	# 25	Compare
Node Classification	Cornell	NLGCN	Accuracy	57.6 ± 5.5	# 48	Compare
Node Classification	Cornell	NLMLP	Accuracy	84.9 ± 5.7	# 21	Compare
Node Classification	Cornell	NLGAT	Accuracy	54.7 ± 7.6	# 51	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	84.9 ± 5.7	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	54.7 ± 7.6	# 27	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	57.6 ± 5.5	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	31.6 ± 1.0	# 24	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	29.5 ± 1.3	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Film(48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	37.9 ± 1.3	# 1	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	88.2 ± 0.5	# 20	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	88.2 ± 0.3	# 20	Compare
Node Classification	PubMed (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	89.0 ± 0.5	# 17	Compare
Node Classification	Squirrel	NLMLP	Accuracy	33.7 ± 1.5	# 48	Compare
Node Classification	Squirrel	NLGCN	Accuracy	59.0 ± 1.2	# 23	Compare
Node Classification	Squirrel	NLGAT	Accuracy	56.8 ± 2.5	# 27	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	33.7 ± 1.5	# 27	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	56.8 ± 2.5	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Squirrel (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	59.0 ± 1.2	# 9	Compare
Node Classification	Texas	NLGAT	Accuracy	62.6 ± 7.1	# 51	Compare
Node Classification	Texas	NLGCN	Accuracy	65.5 ± 6.6	# 49	Compare
Node Classification	Texas	NLMLP	Accuracy	85.4 ± 3.8	# 24	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	62.6 ± 7.1	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	85.4 ± 3.8	# 9	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	65.5 ± 6.6	# 25	Compare
Node Classification	Wisconsin	NLMLP	Accuracy	87.3 ± 4.3	# 23	Compare
Node Classification	Wisconsin	NLGCN	Accuracy	60.2 ± 5.3	# 50	Compare
Node Classification	Wisconsin	NLGAT	Accuracy	56.9 ± 7.3	# 52	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	NLGAT	1:1 Accuracy	56.9 ± 7.3	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	NLGCN	1:1 Accuracy	60.2 ± 5.3	# 25	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin (48%/32%/20% fixed splits)	NLMLP	1:1 Accuracy	87.3 ± 4.3	# 12	Compare

Methods

Add Remove

No methods listed for this paper. Add relevant methods here

Edit Social Preview

Non-Local 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