TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Node Classification	Brazil Air-Traffic	GCN_cheby (Kipf and Welling, 2017)	Accuracy	0.516	# 3
Node Classification	Chameleon (60%/20%/20% random splits)	GCN	1:1 Accuracy	64.18 ± 2.62	# 18
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	GCN	1:1 Accuracy	64.18 ± 2.62	# 17
Node Classification	Citeseer	GCN	Accuracy	70.3	# 60
Node Classification	CiteSeer (60%/20%/20% random splits)	GCN	1:1 Accuracy	81.39 ± 1.23	# 15
Node Classification	Cora	GCN	Accuracy	81.5%	# 61
Node Classification	Cora (60%/20%/20% random splits)	GCN	1:1 Accuracy	87.78 ± 0.96	# 19
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	GCN	1:1 Accuracy	82.46 ± 3.11	# 25
Node Classification	Cornell (60%/20%/20% random splits)	GCN	1:1 Accuracy	82.46 ± 3.11	# 25
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	GCN	1:1 Accuracy	62.23±0.53	# 21
Node Classification	Europe Air-Traffic	GCN (Kipf and Welling, 2017)	Accuracy	37.1	# 6
Node Classification	Europe Air-Traffic	GCN_cheby (Kipf and Welling, 2017)	Accuracy	46.0	# 2
Node Classification	Facebook	GCN (Kipf and Welling, 2017)	Accuracy	57.5	# 4
Node Classification	Facebook	GCN_cheby (Kipf and Welling, 2017)	Accuracy	64.6	# 2
Node Classification	Film (60%/20%/20% random splits)	GCN	1:1 Accuracy	35.51 ± 0.99	# 28
Node Classification	Flickr	GCN_cheby (Kipf and Welling, 2017)	Accuracy	0.479	# 7
Node Classification	Flickr	GCN (Kipf and Welling, 2017)	Accuracy	0.546	# 6
Node Classification	genius	GCN	Accuracy	87.42 ± 0.37	# 14
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	GCN	1:1 Accuracy	87.42 ± 0.37	# 16
Node Classification	NELL	GCN	Accuracy	66.0	# 2
Graph Property Prediction	ogbg-code2	GCN	Test F1 score	0.1507 ± 0.0018	# 18
Graph Property Prediction	ogbg-code2	GCN	Validation F1 score	0.1399 ± 0.0017	# 19
Graph Property Prediction	ogbg-code2	GCN	Number of params	11033210	# 12
Graph Property Prediction	ogbg-code2	GCN	Ext. data	No	# 1
Graph Property Prediction	ogbg-code2	GCN+virtual node	Test F1 score	0.1595 ± 0.0018	# 11
Graph Property Prediction	ogbg-code2	GCN+virtual node	Validation F1 score	0.1461 ± 0.0013	# 12
Graph Property Prediction	ogbg-code2	GCN+virtual node	Number of params	12484310	# 8
Graph Property Prediction	ogbg-code2	GCN+virtual node	Ext. data	No	# 1
Graph Property Prediction	ogbg-molhiv	GCN	Test ROC-AUC	0.7606 ± 0.0097	# 38
Graph Property Prediction	ogbg-molhiv	GCN	Validation ROC-AUC	0.8204 ± 0.0141	# 30
Graph Property Prediction	ogbg-molhiv	GCN	Number of params	527701	# 20
Graph Property Prediction	ogbg-molhiv	GCN	Ext. data	No	# 1
Graph Property Prediction	ogbg-molhiv	GCN (in Julia)	Test ROC-AUC	0.7549 ± 0.0163	# 41
Graph Property Prediction	ogbg-molhiv	GCN (in Julia)	Validation ROC-AUC	0.8042 ± 0.0107	# 35
Graph Property Prediction	ogbg-molhiv	GCN (in Julia)	Number of params	527701	# 20
Graph Property Prediction	ogbg-molhiv	GCN (in Julia)	Ext. data	No	# 1
Graph Property Prediction	ogbg-molhiv	GCN+virtual node	Test ROC-AUC	0.7599 ± 0.0119	# 39
Graph Property Prediction	ogbg-molhiv	GCN+virtual node	Validation ROC-AUC	0.8384 ± 0.0091	# 12
Graph Property Prediction	ogbg-molhiv	GCN+virtual node	Number of params	1978801	# 12
Graph Property Prediction	ogbg-molhiv	GCN+virtual node	Ext. data	No	# 1
Graph Property Prediction	ogbg-molpcba	GCN+virtual node	Test AP	0.2424 ± 0.0034	# 28
Graph Property Prediction	ogbg-molpcba	GCN+virtual node	Validation AP	0.2495 ± 0.0042	# 26
Graph Property Prediction	ogbg-molpcba	GCN+virtual node	Number of params	2017028	# 24
Graph Property Prediction	ogbg-molpcba	GCN+virtual node	Ext. data	No	# 1
Graph Property Prediction	ogbg-molpcba	GCN	Test AP	0.2020 ± 0.0024	# 34
Graph Property Prediction	ogbg-molpcba	GCN	Validation AP	0.2059 ± 0.0033	# 32
Graph Property Prediction	ogbg-molpcba	GCN	Number of params	565928	# 30
Graph Property Prediction	ogbg-molpcba	GCN	Ext. data	No	# 1
Graph Property Prediction	ogbg-ppa	GCN	Test Accuracy	0.6839 ± 0.0084	# 15
Graph Property Prediction	ogbg-ppa	GCN	Validation Accuracy	0.6497 ± 0.0034	# 14
Graph Property Prediction	ogbg-ppa	GCN	Number of params	479437	# 15
Graph Property Prediction	ogbg-ppa	GCN	Ext. data	No	# 1
Graph Property Prediction	ogbg-ppa	GCN+virtual node	Test Accuracy	0.6857 ± 0.0061	# 14
Graph Property Prediction	ogbg-ppa	GCN+virtual node	Validation Accuracy	0.6511 ± 0.0048	# 13
Graph Property Prediction	ogbg-ppa	GCN+virtual node	Number of params	1930537	# 10
Graph Property Prediction	ogbg-ppa	GCN+virtual node	Ext. data	No	# 1
Link Property Prediction	ogbl-citation2	Full-batch GCN	Test MRR	0.8474 ± 0.0021	# 11
Link Property Prediction	ogbl-citation2	Full-batch GCN	Validation MRR	0.8479 ± 0.0023	# 11
Link Property Prediction	ogbl-citation2	Full-batch GCN	Number of params	296449	# 12
Link Property Prediction	ogbl-citation2	Full-batch GCN	Ext. data	No	# 1
Link Property Prediction	ogbl-collab	GCN (val as input)	Test Hits@50	0.4714 ± 0.0145	# 25
Link Property Prediction	ogbl-collab	GCN (val as input)	Validation Hits@50	0.5263 ± 0.0115	# 24
Link Property Prediction	ogbl-collab	GCN (val as input)	Number of params	296449	# 23
Link Property Prediction	ogbl-collab	GCN (val as input)	Ext. data	No	# 1
Link Property Prediction	ogbl-collab	GCN	Test Hits@50	0.4475 ± 0.0107	# 27
Link Property Prediction	ogbl-collab	GCN	Validation Hits@50	0.5263 ± 0.0115	# 24
Link Property Prediction	ogbl-collab	GCN	Number of params	296449	# 23
Link Property Prediction	ogbl-collab	GCN	Ext. data	No	# 1
Link Property Prediction	ogbl-ddi	GCN	Test Hits@20	0.3707 ± 0.0507	# 23
Link Property Prediction	ogbl-ddi	GCN	Validation Hits@20	0.5550 ± 0.0208	# 23
Link Property Prediction	ogbl-ddi	GCN	Number of params	1289985	# 22
Link Property Prediction	ogbl-ddi	GCN	Ext. data	No	# 1
Link Property Prediction	ogbl-ddi	GCN+JKNet	Test Hits@20	0.6056 ± 0.0869	# 19
Link Property Prediction	ogbl-ddi	GCN+JKNet	Validation Hits@20	0.6776 ± 0.0095	# 18
Link Property Prediction	ogbl-ddi	GCN+JKNet	Number of params	1421571	# 18
Link Property Prediction	ogbl-ddi	GCN+JKNet	Ext. data	No	# 1
Link Property Prediction	ogbl-ppa	GCN	Test Hits@100	0.1867 ± 0.0132	# 22
Link Property Prediction	ogbl-ppa	GCN	Validation Hits@100	0.1845 ± 0.0140	# 21
Link Property Prediction	ogbl-ppa	GCN	Number of params	278529	# 14
Link Property Prediction	ogbl-ppa	GCN	Ext. data	No	# 1
Node Property Prediction	ogbn-arxiv	GCN+residual+6 layers	Test Accuracy	0.7286 ± 0.0016	# 51
Node Property Prediction	ogbn-arxiv	GCN+residual+6 layers	Validation Accuracy	0.7382 ± 0.0007	# 54
Node Property Prediction	ogbn-arxiv	GCN+residual+6 layers	Number of params	122542	# 63
Node Property Prediction	ogbn-arxiv	GCN+residual+6 layers	Ext. data	No	# 1
Node Property Prediction	ogbn-arxiv	GCN	Test Accuracy	0.7174 ± 0.0029	# 70
Node Property Prediction	ogbn-arxiv	GCN	Validation Accuracy	0.7300 ± 0.0017	# 66
Node Property Prediction	ogbn-arxiv	GCN	Number of params	110120	# 65
Node Property Prediction	ogbn-arxiv	GCN	Ext. data	No	# 1
Node Property Prediction	ogbn-arxiv	GCN_res + 8 layers	Test Accuracy	0.7262 ± 0.0037	# 56
Node Property Prediction	ogbn-arxiv	GCN_res + 8 layers	Validation Accuracy	0.7369 ± 0.0021	# 55
Node Property Prediction	ogbn-arxiv	GCN_res + 8 layers	Number of params	155824	# 58
Node Property Prediction	ogbn-arxiv	GCN_res + 8 layers	Ext. data	No	# 1
Node Property Prediction	ogbn-arxiv	GCN+residual+node2vec	Test Accuracy	0.7278 ± 0.0013	# 52
Node Property Prediction	ogbn-arxiv	GCN+residual+node2vec	Validation Accuracy	0.7414 ± 0.0008	# 50
Node Property Prediction	ogbn-arxiv	GCN+residual+node2vec	Number of params	21885098	# 7
Node Property Prediction	ogbn-arxiv	GCN+residual+node2vec	Ext. data	No	# 1
Node Property Prediction	ogbn-products	Full-batch GCN	Test Accuracy	0.7564 ± 0.0021	# 55
Node Property Prediction	ogbn-products	Full-batch GCN	Validation Accuracy	0.9200 ± 0.0003	# 42
Node Property Prediction	ogbn-products	Full-batch GCN	Number of params	103727	# 52
Node Property Prediction	ogbn-products	Full-batch GCN	Ext. data	No	# 1
Node Property Prediction	ogbn-proteins	GCN	Test ROC-AUC	0.7251 ± 0.0035	# 22
Node Property Prediction	ogbn-proteins	GCN	Validation ROC-AUC	0.7921 ± 0.0018	# 20
Node Property Prediction	ogbn-proteins	GCN	Number of params	96880	# 21
Node Property Prediction	ogbn-proteins	GCN	Ext. data	No	# 1
Graph Regression	PCQM4Mv2-LSC	GCN	Validation MAE	0.1379	# 17
Graph Regression	PCQM4Mv2-LSC	GCN	Test MAE	0.1398	# 12
Node Classification	Penn94	GCN	Accuracy	82.47 ± 0.27	# 12
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	GCN	1:1 Accuracy	82.47 ± 0.27	# 11
Node Classification	Pubmed	GCN	Accuracy	79.0	# 53
Node Classification	PubMed (60%/20%/20% random splits)	GCN	1:1 Accuracy	88.9 ± 0.32	# 22
Node Classification	Squirrel (60%/20%/20% random splits)	GCN	1:1 Accuracy	44.76 ± 1.39	# 22
Node Classification	Texas (60%/20%/20% random splits)	GCN	1:1 Accuracy	83.11 ± 3.2	# 24
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	GCN	1:1 Accuracy	83.11 ± 3.2	# 22
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	GCN	1:1 Accuracy	62.18 ± 0.26	# 20
Node Classification	Wiki-Vote	GCN_cheby (Kipf and Welling, 2017)	Accuracy	49.5	# 3
Node Classification	Wiki-Vote	GCN (Kipf and Welling, 2017)	Accuracy	32.9	# 5
Node Classification	Wisconsin (60%/20%/20% random splits)	GCN	1:1 Accuracy	75.5 ± 2.92	# 26
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	GCN	1:1 Accuracy	75.5 ± 2.92	# 23

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/graph-property-prediction-on-ogbg-ppa)](https://paperswithcode.com/sota/graph-property-prediction-on-ogbg-ppa?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/link-property-prediction-on-ogbl-citation2)](https://paperswithcode.com/sota/link-property-prediction-on-ogbl-citation2?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/link-property-prediction-on-ogbl-ddi)](https://paperswithcode.com/sota/link-property-prediction-on-ogbl-ddi?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/link-property-prediction-on-ogbl-ppa)](https://paperswithcode.com/sota/link-property-prediction-on-ogbl-ppa?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-property-prediction-on-ogbn-proteins)](https://paperswithcode.com/sota/node-property-prediction-on-ogbn-proteins?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-europe-air-traffic)](https://paperswithcode.com/sota/node-classification-on-europe-air-traffic?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-facebook)](https://paperswithcode.com/sota/node-classification-on-facebook?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-nell)](https://paperswithcode.com/sota/node-classification-on-nell?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-brazil-air-traffic)](https://paperswithcode.com/sota/node-classification-on-brazil-air-traffic?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-wiki-vote)](https://paperswithcode.com/sota/node-classification-on-wiki-vote?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-flickr)](https://paperswithcode.com/sota/node-classification-on-flickr?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/graph-property-prediction-on-ogbg-code2)](https://paperswithcode.com/sota/graph-property-prediction-on-ogbg-code2?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-13)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-13?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-penn94)](https://paperswithcode.com/sota/node-classification-on-penn94?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-genius)](https://paperswithcode.com/sota/node-classification-on-genius?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-citeseer-60-20-20)](https://paperswithcode.com/sota/node-classification-on-citeseer-60-20-20?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-14)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-14?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-4)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-4?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/graph-regression-on-pcqm4mv2-lsc)](https://paperswithcode.com/sota/graph-regression-on-pcqm4mv2-lsc?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-chameleon-60-20-20)](https://paperswithcode.com/sota/node-classification-on-chameleon-60-20-20?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-cora-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-cora-60-20-20-random?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-15)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-15?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-6)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-6?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-pubmed-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-pubmed-60-20-20-random?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-squirrel-60-20-20)](https://paperswithcode.com/sota/node-classification-on-squirrel-60-20-20?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-2)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-2?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic-1)](https://paperswithcode.com/sota/node-classification-on-non-homophilic-1?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-texas-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-texas-60-20-20-random?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-non-homophilic)](https://paperswithcode.com/sota/node-classification-on-non-homophilic?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-cornell-60-20-20)](https://paperswithcode.com/sota/node-classification-on-cornell-60-20-20?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/link-property-prediction-on-ogbl-collab)](https://paperswithcode.com/sota/link-property-prediction-on-ogbl-collab?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-wisconsin-60-20-20)](https://paperswithcode.com/sota/node-classification-on-wisconsin-60-20-20?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-film-60-20-20-random)](https://paperswithcode.com/sota/node-classification-on-film-60-20-20-random?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/graph-property-prediction-on-ogbg-molpcba)](https://paperswithcode.com/sota/graph-property-prediction-on-ogbg-molpcba?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/graph-property-prediction-on-ogbg-molhiv)](https://paperswithcode.com/sota/graph-property-prediction-on-ogbg-molhiv?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-property-prediction-on-ogbn-arxiv)](https://paperswithcode.com/sota/node-property-prediction-on-ogbn-arxiv?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-pubmed)](https://paperswithcode.com/sota/node-classification-on-pubmed?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-property-prediction-on-ogbn-products)](https://paperswithcode.com/sota/node-property-prediction-on-ogbn-products?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-citeseer)](https://paperswithcode.com/sota/node-classification-on-citeseer?p=semi-supervised-classification-with-graph)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/semi-supervised-classification-with-graph/node-classification-on-cora)](https://paperswithcode.com/sota/node-classification-on-cora?p=semi-supervised-classification-with-graph)`

Semi-Supervised Classification with Graph Convolutional Networks

9 Sep 2016 · Thomas N. Kipf, Max Welling ·

We present a scalable approach for semi-supervised learning on graph-structured data that is based on an efficient variant of convolutional neural networks which operate directly on graphs. We motivate the choice of our convolutional architecture via a localized first-order approximation of spectral graph convolutions. Our model scales linearly in the number of graph edges and learns hidden layer representations that encode both local graph structure and features of nodes. In a number of experiments on citation networks and on a knowledge graph dataset we demonstrate that our approach outperforms related methods by a significant margin.

PDF Abstract

Code

Add Remove Mark official

tkipf/pygcn official

5,053

dmlc/dgl

12,994

dmlc/dgl

12,994

tkipf/gcn

7,000

tkipf/keras-gcn

794

See all 51 implementations

Tasks

Add Remove

Document Classification

Drug Discovery

General Classification

Graph Classification

Graph Regression

Node Classification

Node Classification on Non-Homophilic (Heterophilic) Graphs

Node Property Prediction

Skeleton Based Action Recognition

Datasets

Pubmed

OGB

Flickr30k

Cora Citeseer

NELL Penn94 genius twitch-gamers Deezer-Europe Wisconsin(60%/20%/20% random splits) Squirrel (60%/20%/20% random splits) Texas(60%/20%/20% random splits) Film (60%/20%/20% random splits) Cornell (60%/20%/20% random splits) PubMed (60%/20%/20% random splits) Chameleon(60%/20%/20% random splits)

PCQM4Mv2-LSC Brazil Air-Traffic

Drug Combination Extraction Dataset

Results from the Paper

Edit

Ranked #1 on Graph Property Prediction on ogbg-ppa

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Node Classification	Chameleon (60%/20%/20% random splits)	GCN	1:1 Accuracy	64.18 ± 2.62	# 18	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Chameleon(60%/20%/20% random splits)	GCN	1:1 Accuracy	64.18 ± 2.62	# 17	Compare
Node Classification	Citeseer	GCN	Accuracy	70.3	# 60	Compare
Node Classification	CiteSeer (60%/20%/20% random splits)	GCN	1:1 Accuracy	81.39 ± 1.23	# 15	Compare
Node Classification	Cora	GCN	Accuracy	81.5%	# 61	Compare
Node Classification	Cora (60%/20%/20% random splits)	GCN	1:1 Accuracy	87.78 ± 0.96	# 19	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Cornell (60%/20%/20% random splits)	GCN	1:1 Accuracy	82.46 ± 3.11	# 25	Compare
Node Classification	Cornell (60%/20%/20% random splits)	GCN	1:1 Accuracy	82.46 ± 3.11	# 25	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Deezer-Europe	GCN	1:1 Accuracy	62.23±0.53	# 21	Compare
Node Classification	Film (60%/20%/20% random splits)	GCN	1:1 Accuracy	35.51 ± 0.99	# 28	Compare
Node Classification	Flickr	GCN_cheby (Kipf and Welling, 2017)	Accuracy	0.479	# 7	Compare
Node Classification	Flickr	GCN (Kipf and Welling, 2017)	Accuracy	0.546	# 6	Compare
Node Classification	genius	GCN	Accuracy	87.42 ± 0.37	# 14	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	genius	GCN	1:1 Accuracy	87.42 ± 0.37	# 16	Compare
Node Classification	NELL	GCN	Accuracy	66.0	# 2	Compare
Graph Property Prediction	ogbg-code2	GCN	Test F1 score	0.1507 ± 0.0018	# 18	Compare
			Validation F1 score	0.1399 ± 0.0017	# 19	Compare
			Number of params	11033210	# 12	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-code2	GCN+virtual node	Test F1 score	0.1595 ± 0.0018	# 11	Compare
			Validation F1 score	0.1461 ± 0.0013	# 12	Compare
			Number of params	12484310	# 8	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-molhiv	GCN	Test ROC-AUC	0.7606 ± 0.0097	# 38	Compare
			Validation ROC-AUC	0.8204 ± 0.0141	# 30	Compare
			Number of params	527701	# 20	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-molhiv	GCN (in Julia)	Test ROC-AUC	0.7549 ± 0.0163	# 41	Compare
			Validation ROC-AUC	0.8042 ± 0.0107	# 35	Compare
			Number of params	527701	# 20	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-molhiv	GCN+virtual node	Test ROC-AUC	0.7599 ± 0.0119	# 39	Compare
			Validation ROC-AUC	0.8384 ± 0.0091	# 12	Compare
			Number of params	1978801	# 12	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-molpcba	GCN+virtual node	Test AP	0.2424 ± 0.0034	# 28	Compare
			Validation AP	0.2495 ± 0.0042	# 26	Compare
			Number of params	2017028	# 24	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-molpcba	GCN	Test AP	0.2020 ± 0.0024	# 34	Compare
			Validation AP	0.2059 ± 0.0033	# 32	Compare
			Number of params	565928	# 30	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-ppa	GCN	Test Accuracy	0.6839 ± 0.0084	# 15	Compare
			Validation Accuracy	0.6497 ± 0.0034	# 14	Compare
			Number of params	479437	# 15	Compare
			Ext. data	No	# 1	Compare
Graph Property Prediction	ogbg-ppa	GCN+virtual node	Test Accuracy	0.6857 ± 0.0061	# 14	Compare
			Validation Accuracy	0.6511 ± 0.0048	# 13	Compare
			Number of params	1930537	# 10	Compare
			Ext. data	No	# 1	Compare
Link Property Prediction	ogbl-citation2	Full-batch GCN	Test MRR	0.8474 ± 0.0021	# 11	Compare
			Validation MRR	0.8479 ± 0.0023	# 11	Compare
			Number of params	296449	# 12	Compare
			Ext. data	No	# 1	Compare
Link Property Prediction	ogbl-collab	GCN (val as input)	Test Hits@50	0.4714 ± 0.0145	# 25	Compare
			Validation Hits@50	0.5263 ± 0.0115	# 24	Compare
			Number of params	296449	# 23	Compare
			Ext. data	No	# 1	Compare
Link Property Prediction	ogbl-collab	GCN	Test Hits@50	0.4475 ± 0.0107	# 27	Compare
			Validation Hits@50	0.5263 ± 0.0115	# 24	Compare
			Number of params	296449	# 23	Compare
			Ext. data	No	# 1	Compare
Link Property Prediction	ogbl-ddi	GCN	Test Hits@20	0.3707 ± 0.0507	# 23	Compare
			Validation Hits@20	0.5550 ± 0.0208	# 23	Compare
			Number of params	1289985	# 22	Compare
			Ext. data	No	# 1	Compare
Link Property Prediction	ogbl-ddi	GCN+JKNet	Test Hits@20	0.6056 ± 0.0869	# 19	Compare
			Validation Hits@20	0.6776 ± 0.0095	# 18	Compare
			Number of params	1421571	# 18	Compare
			Ext. data	No	# 1	Compare
Link Property Prediction	ogbl-ppa	GCN	Test Hits@100	0.1867 ± 0.0132	# 22	Compare
			Validation Hits@100	0.1845 ± 0.0140	# 21	Compare
			Number of params	278529	# 14	Compare
			Ext. data	No	# 1	Compare
Node Property Prediction	ogbn-arxiv	GCN+residual+6 layers	Test Accuracy	0.7286 ± 0.0016	# 51	Compare
			Validation Accuracy	0.7382 ± 0.0007	# 54	Compare
			Number of params	122542	# 63	Compare
			Ext. data	No	# 1	Compare
Node Property Prediction	ogbn-arxiv	GCN	Test Accuracy	0.7174 ± 0.0029	# 70	Compare
			Validation Accuracy	0.7300 ± 0.0017	# 66	Compare
			Number of params	110120	# 65	Compare
			Ext. data	No	# 1	Compare
Node Property Prediction	ogbn-arxiv	GCN_res + 8 layers	Test Accuracy	0.7262 ± 0.0037	# 56	Compare
			Validation Accuracy	0.7369 ± 0.0021	# 55	Compare
			Number of params	155824	# 58	Compare
			Ext. data	No	# 1	Compare
Node Property Prediction	ogbn-arxiv	GCN+residual+node2vec	Test Accuracy	0.7278 ± 0.0013	# 52	Compare
			Validation Accuracy	0.7414 ± 0.0008	# 50	Compare
			Number of params	21885098	# 7	Compare
			Ext. data	No	# 1	Compare
Node Property Prediction	ogbn-products	Full-batch GCN	Test Accuracy	0.7564 ± 0.0021	# 55	Compare
			Validation Accuracy	0.9200 ± 0.0003	# 42	Compare
			Number of params	103727	# 52	Compare
			Ext. data	No	# 1	Compare
Node Property Prediction	ogbn-proteins	GCN	Test ROC-AUC	0.7251 ± 0.0035	# 22	Compare
			Validation ROC-AUC	0.7921 ± 0.0018	# 20	Compare
			Number of params	96880	# 21	Compare
			Ext. data	No	# 1	Compare
Graph Regression	PCQM4Mv2-LSC	GCN	Validation MAE	0.1379	# 17	Compare
Graph Regression	PCQM4Mv2-LSC	GCN	Test MAE	0.1398	# 12	Compare
Node Classification	Penn94	GCN	Accuracy	82.47 ± 0.27	# 12	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Penn94	GCN	1:1 Accuracy	82.47 ± 0.27	# 11	Compare
Node Classification	Pubmed	GCN	Accuracy	79.0	# 53	Compare
Node Classification	PubMed (60%/20%/20% random splits)	GCN	1:1 Accuracy	88.9 ± 0.32	# 22	Compare
Node Classification	Squirrel (60%/20%/20% random splits)	GCN	1:1 Accuracy	44.76 ± 1.39	# 22	Compare
Node Classification	Texas (60%/20%/20% random splits)	GCN	1:1 Accuracy	83.11 ± 3.2	# 24	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Texas(60%/20%/20% random splits)	GCN	1:1 Accuracy	83.11 ± 3.2	# 22	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	twitch-gamers	GCN	1:1 Accuracy	62.18 ± 0.26	# 20	Compare
Node Classification	Wisconsin (60%/20%/20% random splits)	GCN	1:1 Accuracy	75.5 ± 2.92	# 26	Compare
Node Classification on Non-Homophilic (Heterophilic) Graphs	Wisconsin(60%/20%/20% random splits)	GCN	1:1 Accuracy	75.5 ± 2.92	# 23	Compare

Results from Other Papers

Task	Dataset	Model	Metric Name	Metric Value	Rank	Compare
Node Classification	Brazil Air-Traffic	GCN_cheby (Kipf and Welling, 2017)	Accuracy	0.516	# 3	See all
Node Classification	Europe Air-Traffic	GCN (Kipf and Welling, 2017)	Accuracy	37.1	# 6	See all
Node Classification	Europe Air-Traffic	GCN_cheby (Kipf and Welling, 2017)	Accuracy	46.0	# 2	See all
Node Classification	Facebook	GCN (Kipf and Welling, 2017)	Accuracy	57.5	# 4	See all
Node Classification	Facebook	GCN_cheby (Kipf and Welling, 2017)	Accuracy	64.6	# 2	See all
Node Classification	Wiki-Vote	GCN_cheby (Kipf and Welling, 2017)	Accuracy	49.5	# 3	See all
Node Classification	Wiki-Vote	GCN (Kipf and Welling, 2017)	Accuracy	32.9	# 5	See all

Methods

Add Remove

GCN • Graph Convolutional Networks

Edit Social Preview

Semi-Supervised Classification with Graph Convolutional Networks

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit