CORE: Contrastive Masked Feature Reconstruction on Graphs

Jianyuan Bo, Yuan Fang

Abstract

In the rapidly evolving field of self-supervised learning on graphs, generative and contrastive methodologies have emerged as two dominant approaches. Our study focuses on masked feature reconstruction (MFR), a generative technique where a model learns to restore the raw features of masked nodes in a self-supervised manner. We observe that both MFR and graph contrastive learning (GCL) aim to maximize agreement between similar elements. Building on this observation, we reveal a novel theoretical insight: under specific conditions, the objectives of MFR and node-level GCL converge, despite their distinct operational mechanisms. This theoretical connection suggests these approaches are complementary rather than fundamentally different, prompting us to explore their integration to enhance self-supervised learning on graphs. Our research presents Contrastive Masked Feature Reconstruction (CORE), a novel graph self-supervised learning framework that integrates contrastive learning into MFR. Specifically, we form positive pairs exclusively between the original and reconstructed features of masked nodes, encouraging the encoder to prioritize contextual information over the node's own features. Additionally, we leverage the masked nodes themselves as negative samples, combining MFR's reconstructive power with GCL's discriminative ability to better capture intrinsic graph structures. Empirically, our proposed framework CORE significantly outperforms MFR across node and graph classification tasks, demonstrating state-of-the-art results. In particular, CORE surpasses GraphMAE and GraphMAE2 by up to 2.80\

Benchmark Results

Tested on 9 datasets 41 baseline values reported too low

Tables show presented results in global context (best 3 papers + GCN + MLP).

Classic

#	Model	Accuracy ↑	Paper
1	GraphTARIF code ↗ Hyb global	0.9461 ±0.0017	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
2	GCN GNN baseline actual #2 vs paper #270 268 places worse	0.9412 ±0.0008 best reported	actual #2 vs paper #270 268 places worse paper 0.8650 ±0.0050 best at pub: Diss-l-ECT: Dissecting Graph Data with local Euler Characteristic Transforms (ICML 2024) best today: Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification (2026) paper 0.8650 ±0.0050 Δ at pub 0.0510 Δ today 0.0762
3	GAT GNN baseline actual #3 vs paper #256 253 places worse	0.9398 ±0.0022 best reported	actual #3 vs paper #256 253 places worse paper 0.8690 ±0.0030 best at pub: Diss-l-ECT: Dissecting Graph Data with local Euler Characteristic Transforms (ICML 2024) best today: Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification (2026) paper 0.8690 ±0.0030 Δ at pub 0.0550 Δ today 0.0708
8	SAGE code ↗ GNN global	0.9351 ±0.0006	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
56	CORE Hyb ★ proposed	0.9100 ±0.0030	This paper
79	BGRL GNN baseline actual #79 vs paper #81 2 places worse	0.9034 ±0.0019 best reported	actual #79 vs paper #81 2 places worse paper 0.9030 ±0.0020 best at pub: Spectral Augmentation for Self-Supervised Learning on Graphs (ICLR 2022) best today: Adaptive Negative Scheduling for Graph Contrastive Learning (2026) Difference within reported uncertainty; not counted as underperformance. paper 0.9030 ±0.0020 Δ at pub 0.0004 Δ today 0.0004
84	ABGML GNN baseline	0.9020 ±0.0030	This paper
100	GraphMAE GNN baseline actual #100 vs paper #130 30 places worse	0.9002 ±0.0024 best reported	actual #100 vs paper #130 30 places worse paper 0.8960 ±0.0030 best at pub: Discrepancy-Aware Graph Mask Auto-Encoder (KDD 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8960 ±0.0030 Δ at pub 0.0042
111	GraphMAE2 GNN baseline actual #111 vs paper #111	0.8990 ±0.0030	actual #111 vs paper #111 paper 0.8990 ±0.0030 prev best: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) Improvement within reported uncertainty; not highlighted. paper 0.8990 ±0.0030 Δ at pub 0.0030
139	GRACE GNN baseline actual #139 vs paper #139	0.8953 ±0.0035 best reported	actual #139 vs paper #139 paper 0.8950 ±0.0040 best at pub: Spectral Augmentation for Self-Supervised Learning on Graphs (ICLR 2022) best today: Adaptive Negative Scheduling for Graph Contrastive Learning (2026) Difference within reported uncertainty; not counted as underperformance. paper 0.8950 ±0.0040 Δ at pub 0.0003 Δ today 0.0003
159	CCA-SSG GNN baseline actual #159 vs paper #171 12 places worse	0.8896 ±0.0013 best reported	actual #159 vs paper #171 12 places worse paper 0.8870 ±0.0030 best at pub: Str-GCL: Structural Commonsense Driven Graph Contrastive Learning (WWW 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8870 ±0.0030 Δ at pub 0.0026
160	GCA GNN baseline actual #160 vs paper #212 52 places worse	0.8894 ±0.0015 best reported	actual #160 vs paper #212 52 places worse paper 0.8790 ±0.0030 best at pub: Barlow Graph Auto-Encoder for Unsupervised Network Embedding (2021) best today: Adaptive Negative Scheduling for Graph Contrastive Learning (2026) paper 0.8790 ±0.0030 Δ at pub 0.0104 Δ today 0.0104
179	MVGRL GNN baseline actual #179 vs paper #235 56 places worse	0.8861 ±0.0064 best reported	actual #179 vs paper #235 56 places worse paper 0.8750 ±0.0010 best at pub: Balancing Graph Embedding Smoothness in Self-Supervised Learning via Information-Theoretic Decomposition (WWW 2025) paper 0.8750 ±0.0010 Δ at pub 0.0111
210	DGI GNN baseline actual #210 vs paper #350 140 places worse	0.8798 ±0.0081 best reported	actual #210 vs paper #350 140 places worse paper 0.8400 ±0.0050 best at pub: Bootstrap Latents of Nodes and Neighbors for Graph Self-Supervised Learning (2024) paper 0.8400 ±0.0050 Δ at pub 0.0398
264	GAE GNN baseline actual #264 vs paper #303 39 places worse	0.8666 ±0.0007 best reported	actual #264 vs paper #303 39 places worse paper 0.8530 ±0.0020 best at pub: Graph Masked Autoencoders with Transformers (2022) paper 0.8530 ±0.0020 Δ at pub 0.0136
320	MLP code ↗ Trad global	0.8463	GraphFM: A generalist graph transformer that learns transferable representations across diverse … code ↗
342	GMI GNN baseline actual #342 vs paper #398 56 places worse	0.8422 ±0.0052 best reported	actual #342 vs paper #398 56 places worse paper 0.8220 ±0.0030 best at pub: Localized Graph Contrastive Learning (2022) paper 0.8220 ±0.0030 Δ at pub 0.0202

Coauthor-Physics

Global leaderboard →

#	Model	Accuracy ↑	Paper
1	GrokFormer code ↗ GT global	0.9831 ±0.0018	GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers code ↗
2	PolyFormer code ↗ GT global	0.9808 ±0.0027	GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers code ↗
3	SGFormer code ↗ GT global	0.9796 ±0.0081	GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers code ↗
6	GAT GNN baseline actual #6 vs paper #116 110 places worse	0.9782 ±0.0028 best reported	actual #6 vs paper #116 110 places worse paper 0.9550 ±0.0020 best at pub: GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers (ICML 2024) paper 0.9550 ±0.0020 Δ at pub 0.0232
9	GCN GNN baseline actual #9 vs paper #84 75 places worse	0.9774 ±0.0035 best reported	actual #9 vs paper #84 75 places worse paper 0.9570 ±0.0020 best at pub: GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers (ICML 2024) paper 0.9570 ±0.0020 Δ at pub 0.0204
17	SAGE code ↗ GNN global	0.9739 ±0.0018	Clarify Confused Nodes via Separated Learning code ↗
53	ABGML GNN baseline	0.9610 ±0.0010	This paper
58	MLP code ↗ Trad global	0.9597 ±0.0007	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
63	GCA GNN baseline actual #63 vs paper #84 21 places worse	0.9590 ±0.0020 best reported	actual #63 vs paper #84 21 places worse paper 0.9570 best at pub: Unifying Graph Contrastive Learning with Flexible Contextual Scopes (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.9570 Δ at pub 0.0020
69	BGRL GNN baseline actual #69 vs paper #84 15 places worse	0.9580 ±0.0040 best reported	actual #69 vs paper #84 15 places worse paper 0.9570 best at pub: Unifying Graph Contrastive Learning with Flexible Contextual Scopes (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.9570 Δ at pub 0.0010
105	CORE Hyb ★ proposed	0.9560 ±0.0010	This paper
106	CCA-SSG GNN baseline actual #106 vs paper #125 19 places worse	0.9559 ±0.0007 best reported	actual #106 vs paper #125 19 places worse paper 0.9540 ±0.0010 best at pub: Wiener Graph Deconvolutional Network Improves Graph Self-Supervised Learning (AAAI 2022) paper 0.9540 ±0.0010 Δ at pub 0.0019
117	GraphMAE GNN baseline actual #117 vs paper #117	0.9550 ±0.0010	actual #117 vs paper #117 paper 0.9550 ±0.0010 prev best: Rethinking and Simplifying Bootstrapped Graph Latents (2023) Improvement within reported uncertainty; not highlighted. paper 0.9550 ±0.0010 Δ at pub 0.0010
118	GraphMAE2 GNN baseline actual #118 vs paper #118	0.9550 ±0.0020	actual #118 vs paper #118 paper 0.9550 ±0.0020 prev best: Balancing Graph Embedding Smoothness in Self-Supervised Learning via Information-Theoretic Decomposition (WWW 2025) Improvement within reported uncertainty; not highlighted. paper 0.9550 ±0.0020 Δ at pub 0.0025
125	MVGRL GNN baseline actual #125 vs paper #136 11 places worse	0.9544 ±0.0027 best reported	actual #125 vs paper #136 11 places worse paper 0.9530 best at pub: HAT-GAE: Self-Supervised Graph Auto-encoders with Hierarchical Adaptive Masking and Trainable Corruption (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.9530 Δ at pub 0.0014
146	GAE GNN baseline actual #146 vs paper #161 15 places worse	0.9520 ±0.0010 best reported	actual #146 vs paper #161 15 places worse paper 0.9490 ±0.0010 best at pub: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) paper 0.9490 ±0.0010 Δ at pub 0.0030
167	DGI GNN baseline actual #167 vs paper #174 7 places worse	0.9489 ±0.0009 best reported	actual #167 vs paper #174 7 places worse paper 0.9450 ±0.0050 best at pub: Balancing Graph Embedding Smoothness in Self-Supervised Learning via Information-Theoretic Decomposition (WWW 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.9450 ±0.0050 Δ at pub 0.0039

Coauthor-CS

Global leaderboard →

#	Model	Accuracy ↑	Paper
1	DoG code ↗ GNN global	0.9690 ±0.0030	Diffusion on Graph: Augmentation of Graph Structure for Node Classification code ↗
2	GWN-fa code ↗ Hyb global	0.9667 ±0.0026	Graph Wave Networks code ↗
3	GWN-sym code ↗ GNN global	0.9666 ±0.0026	Graph Wave Networks code ↗
6	GraphSAGE code ↗ GNN global	0.9593 ±0.0025	Graph Wave Networks code ↗
12	GCN GNN baseline actual #12 vs paper #69 57 places worse	0.9514 ±0.0025 best reported	actual #12 vs paper #69 57 places worse paper 0.9300 ±0.0030 best at pub: Graph Wave Networks (WWW 2025) paper 0.9300 ±0.0030 Δ at pub 0.0214
21	GAT GNN baseline actual #21 vs paper #107 86 places worse	0.9437 ±0.0028 best reported	actual #21 vs paper #107 86 places worse paper 0.9230 ±0.0020 best at pub: Graph Wave Networks (WWW 2025) paper 0.9230 ±0.0020 Δ at pub 0.0207
38	ABGML GNN baseline	0.9360 ±0.0020	This paper
43	DGI GNN baseline actual #43 vs paper #113 70 places worse	0.9350 ±0.0030 best reported	actual #43 vs paper #113 70 places worse paper 0.9220 ±0.0060 best at pub: PyG-SSL: A Graph Self-Supervised Learning Toolkit (2024) paper 0.9220 ±0.0060 Δ at pub 0.0130
48	CCA-SSG GNN baseline actual #48 vs paper #54 6 places worse	0.9337 ±0.0020 best reported	actual #48 vs paper #54 6 places worse paper 0.9330 ±0.0020 best at pub: MA-GCL: Model Augmentation Tricks for Graph Contrastive Learning (AAAI 2022) Difference within reported uncertainty; not counted as underperformance. paper 0.9330 ±0.0020 Δ at pub 0.0007
51	GCA GNN baseline actual #51 vs paper #63 12 places worse	0.9332 ±0.0012 best reported	actual #51 vs paper #63 12 places worse paper 0.9310 best at pub: Beyond Smoothing: Unsupervised Graph Representation Learning with Edge Heterophily Discriminating (AAAI 2022) paper 0.9310 Δ at pub 0.0022
53	BGRL GNN baseline actual #53 vs paper #54 1 place worse	0.9331 ±0.0013 best reported	actual #53 vs paper #54 1 place worse paper 0.9330 ±0.0010 best at pub: Spectral Augmentation for Self-Supervised Learning on Graphs (ICLR 2022) best today: Adaptive Negative Scheduling for Graph Contrastive Learning (2026) Difference within reported uncertainty; not counted as underperformance. paper 0.9330 ±0.0010 Δ at pub 0.0001 Δ today 0.0001
61	CORE Hyb ★ proposed	0.9320 ±0.0020	This paper
63	GraphMAE GNN baseline actual #63 vs paper #82 19 places worse	0.9313 ±0.0015 best reported	actual #63 vs paper #82 19 places worse paper 0.9290 ±0.0020 best at pub: Rethinking Graph Contrastive Learning through Relative Similarity Preservation (IJCAI 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.9290 ±0.0020 Δ at pub 0.0023
66	GraphMAE2 GNN baseline	0.9310 ±0.0020	This paper
69	GRACE GNN baseline actual #69 vs paper #148 79 places worse	0.9301 ±0.0003 best reported	actual #69 vs paper #148 79 places worse paper 0.9110 ±0.0020 best at pub: A Fused Gromov-Wasserstein Approach to Subgraph Contrastive Learning (TMLR 2025) paper 0.9110 ±0.0020 Δ at pub 0.0191
88	MLP Trad global	0.9280 ±0.0020	Revisiting the Necessity of Graph Learning and Common Graph Benchmarks
116	MVGRL GNN baseline actual #116 vs paper #118 2 places worse	0.9211 ±0.0012 best reported	actual #116 vs paper #118 2 places worse paper 0.9210 ±0.0010 best at pub: Graph Contrastive Learning with Adaptive Augmentation (WWW 2020) best today: Adaptive Negative Scheduling for Graph Contrastive Learning (2026) Difference within reported uncertainty; not counted as underperformance. paper 0.9210 ±0.0010 Δ at pub 0.0001 Δ today 0.0001
163	GAE GNN baseline actual #163 vs paper #163	0.9005 ±0.0035 best reported	actual #163 vs paper #163 paper 0.9000 ±0.0070 best at pub: HAT-GAE: Self-Supervised Graph Auto-encoders with Hierarchical Adaptive Masking and Trainable Corruption (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.9000 ±0.0070 Δ at pub 0.0005

TU Dortmund

MUTAG

Global leaderboard →

#	Model	Accuracy ↑	Paper
1	MSH-GNN GNN global	0.9910 ±0.0030	Multi-Scale Harmonic Encoding for Feature-Wise Graph Message Passing
2	ID-GNN+BEC Hyb global	0.9830 ±0.0360	Depth-Adaptive Graph Neural Networks via Learnable Bakry-Émery Curvature
3	SupCosine Hyb global	0.9830 ±0.0250	Supervised Contrastive Learning with Structure Inference for Graph Classification
39	GIN GNN baseline actual #39 vs paper #237 198 places worse	0.9330 ±0.0290 best reported	actual #39 vs paper #237 198 places worse paper 0.8940 ±0.0560 best at pub: ProtGNN: Towards Self-Explaining Graph Neural Networks (AAAI 2021) Difference within reported uncertainty; not counted as underperformance. paper 0.8940 ±0.0560 Δ at pub 0.0390
63	GCN GNN global	0.9220 ±0.0440	Depth-Adaptive Graph Neural Networks via Learnable Bakry-Émery Curvature
102	GCMAE GNN baseline actual #102 vs paper #102	0.9130 ±0.0060	actual #102 vs paper #102 paper 0.9130 ±0.0060 prev best: Generative and Contrastive Paradigms Are Complementary for Graph Self-Supervised Learning (2023) Improvement within reported uncertainty; not highlighted. paper 0.9130 ±0.0060 Δ at pub 0.0002
105	InfoGCL GNN baseline	0.9120 ±0.0130	This paper
159	WL Trad baseline actual #159 vs paper #597 438 places worse	0.9040 ±0.0570 best reported	actual #159 vs paper #597 438 places worse paper 0.8070 ±0.0300 best at pub: How Powerful are $K$-hop Message Passing Graph Neural Networks (NeurIPS 2022) paper 0.8070 ±0.0300 Δ at pub 0.0970
186	GraphMAE2 + CORE GNN ★ proposed	0.9010 ±0.0110	This paper
199	MVGRL GNN baseline actual #199 vs paper #213 14 places worse	0.8993 ±0.0616 best reported	actual #199 vs paper #213 14 places worse paper 0.8970 ±0.0110 best at pub: Improving Expressivity of GNNs with Subgraph-specific Factor Embedded Normalization (KDD 2023) Difference within reported uncertainty; not counted as underperformance. paper 0.8970 ±0.0110 Δ at pub 0.0023
210	AUG-MAE GNN baseline actual #210 vs paper #318 108 places worse	0.8980 ±0.0130 best reported	actual #210 vs paper #318 108 places worse paper 0.8830 ±0.0100 best at pub: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.8830 ±0.0100 Δ at pub 0.0150
226	GraphMAE2 GNN baseline actual #226 vs paper #226	0.8960 ±0.0030	actual #226 vs paper #226 paper 0.8960 ±0.0030 prev best: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) Improvement within reported uncertainty; not highlighted. paper 0.8960 ±0.0030 Δ at pub 0.0010
254	GraphMAE + CORE GNN ★ proposed	0.8940 ±0.0050	This paper
239	DiffPool GNN baseline actual #239 vs paper #447 208 places worse	0.8940 ±0.0460 best reported	actual #239 vs paper #447 208 places worse paper 0.8500 ±0.1030 best at pub: Graph Classification via Reference Distribution Learning: Theory and Practice (NeurIPS 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.8500 ±0.1030 Δ at pub 0.0440
257	InfoGraph GNN baseline actual #257 vs paper #272 15 places worse	0.8930 ±0.0514 best reported	actual #257 vs paper #272 15 places worse paper 0.8900 ±0.0110 best at pub: Improving Expressivity of GNNs with Subgraph-specific Factor Embedded Normalization (KDD 2023) Difference within reported uncertainty; not counted as underperformance. paper 0.8900 ±0.0110 Δ at pub 0.0030
258	GraphMAE GNN baseline actual #258 vs paper #325 67 places worse	0.8930 ±0.0110 best reported	actual #258 vs paper #325 67 places worse paper 0.8820 ±0.0130 best at pub: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.8820 ±0.0130 Δ at pub 0.0110
351	JOAO GNN baseline actual #351 vs paper #366 15 places worse	0.8770 ±0.0080 best reported	actual #351 vs paper #366 15 places worse paper 0.8740 ±0.0100 best at pub: Graph Self-Supervised Learning: A Survey (2021) best today: Learning Graph Augmentations to Learn Graph Representations (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.8740 ±0.0100 Δ at pub 0.0030 Δ today 0.0030
354	GraphCL GNN baseline actual #354 vs paper #396 42 places worse	0.8766 ±0.0103 best reported	actual #354 vs paper #396 42 places worse paper 0.8680 ±0.0130 best at pub: Graph Masked Autoencoders with Transformers (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.8680 ±0.0130 Δ at pub 0.0086
365	DGK SVM baseline actual #365 vs paper #366 1 place worse	0.8744 ±0.0272 best reported	actual #365 vs paper #366 1 place worse paper 0.8740 ±0.0270 best at pub: Provably Powerful Graph Networks (NeurIPS 2019) best today: Model-Driven Graph Contrastive Learning (2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8740 ±0.0270 Δ at pub 0.0004 Δ today 0.0004
403	MLP code ↗ Trad global	0.8660 ±0.0495	Breaking the Limits of Message Passing Graph Neural Networks code ↗
407	graph2vec Walk baseline actual #407 vs paper #515 108 places worse	0.8644 ±0.0125 best reported	actual #407 vs paper #515 108 places worse paper 0.8320 ±0.0920 best at pub: Graph Masked Autoencoders with Transformers (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.8320 ±0.0920 Δ at pub 0.0324

NCI1

Global leaderboard →

#	Model	Accuracy ↑	Paper
1	WL-MLP GT global	0.9954	1-WL Expressiveness Is (Almost) All You Need
2	GIN GNN baseline actual #2 vs paper #91 89 places worse	0.9934 best reported	actual #2 vs paper #91 89 places worse paper 0.8270 ±0.0170 best at pub: 1-WL Expressiveness Is (Almost) All You Need (2022) paper 0.8270 ±0.0170 Δ at pub 0.1664
3	GCN GNN global	0.9727	1-WL Expressiveness Is (Almost) All You Need
12	WL Trad baseline actual #12 vs paper #211 199 places worse	0.8590 ±0.0010 best reported	actual #12 vs paper #211 199 places worse paper 0.8030 ±0.0050 best at pub: Propagation Kernels (2014) paper 0.8030 ±0.0050 Δ at pub 0.0560
67	CORE (GraphMAE + CORE) ★ proposed	0.8320 ±0.0020	This paper
104	GraphCL GNN baseline actual #104 vs paper #325 221 places worse	0.8250 ±0.0013 best reported	actual #104 vs paper #325 221 places worse paper 0.7790 ±0.0040 best at pub: Label-invariant Augmentation for Semi-Supervised Graph Classification (NeurIPS 2022) paper 0.7790 ±0.0040 Δ at pub 0.0460
116	GraphMAE2 GNN baseline actual #116 vs paper #140 24 places worse	0.8230 ±0.0090 best reported	actual #116 vs paper #140 24 places worse paper 0.8180 ±0.0050 best at pub: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.8180 ±0.0050 Δ at pub 0.0050
131	GraphSAGE code ↗ GNN global	0.8200 ±0.0100	Weisfeiler and Leman Go Infinite: Spectral and Combinatorial Pre-Colorings code ↗
155	GCMAE GNN baseline actual #155 vs paper #155	0.8142 ±0.0030 best reported	actual #155 vs paper #155 paper 0.8140 ±0.0030 best at pub: Generative and Contrastive Paradigms Are Complementary for Graph Self-Supervised Learning (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.8140 ±0.0030 Δ at pub 0.0002
200	GraphMAE GNN baseline actual #200 vs paper #202 2 places worse	0.8042 ±0.0035 best reported	actual #200 vs paper #202 2 places worse paper 0.8040 ±0.0030 best at pub: Self-Supervised Graph Learning via Spectral Bootstrapping and Laplacian-Based Augmentations (2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8040 ±0.0030 Δ at pub 0.0002
210	DGK SVM baseline actual #210 vs paper #211 1 place worse	0.8031 ±0.0046 best reported	actual #210 vs paper #211 1 place worse paper 0.8030 ±0.0050 best at pub: Graph Capsule Convolutional Neural Networks (ICML 2018) best today: Model-Driven Graph Contrastive Learning (2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8030 ±0.0050 Δ at pub 0.0001 Δ today 0.0001
215	InfoGCL GNN baseline	0.8020 ±0.0060	This paper
301	JOAO GNN baseline actual #301 vs paper #315 14 places worse	0.7840 ±0.0050 best reported	actual #301 vs paper #315 14 places worse paper 0.7810 ±0.0050 best at pub: Omni-Granular Ego-Semantic Propagation for Self-Supervised Graph Representation Learning (ICML 2022) Difference within reported uncertainty; not counted as underperformance. paper 0.7810 ±0.0050 Δ at pub 0.0030
338	InfoGraph GNN baseline actual #338 vs paper #386 48 places worse	0.7750 ±0.0074 best reported	actual #338 vs paper #386 48 places worse paper 0.7620 ±0.0110 best at pub: Graph Masked Autoencoders with Transformers (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.7620 ±0.0110 Δ at pub 0.0130
406	graph2vec Walk baseline actual #406 vs paper #446 40 places worse	0.7540 ±0.0120 best reported	actual #406 vs paper #446 40 places worse paper 0.7320 ±0.0180 best at pub: InfoGCL: Information-Aware Graph Contrastive Learning (NeurIPS 2021) Difference within reported uncertainty; not counted as underperformance. paper 0.7320 ±0.0180 Δ at pub 0.0220

COLLAB

Global leaderboard →

#	Model	Accuracy ↑	Paper
1	WL-MLP GT global	0.9792	1-WL Expressiveness Is (Almost) All You Need
2	MSH-GNN GNN global	0.9640 ±0.0070	Multi-Scale Harmonic Encoding for Feature-Wise Graph Message Passing
3	CoCN code ↗ GNN global	0.8722 ±0.0013	Scalable Graph Compressed Convolutions code ↗
9	CORE (GraphMAE2 + CORE) ★ proposed	0.8430 ±0.0010	This paper
10	GraphMAE2 GNN baseline actual #10 vs paper #10	0.8410 ±0.0030	actual #10 vs paper #10 paper 0.8410 ±0.0030 prev best: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) paper 0.8410 ±0.0030 Δ at pub 0.0240
22	GraphMAE GNN baseline actual #22 vs paper #109 87 places worse	0.8270 ±0.0100 best reported	actual #22 vs paper #109 87 places worse paper 0.8030 ±0.0050 best at pub: Revisiting Graph Autoencoders as Implicit Contrastive Learners (KDD 2024) paper 0.8030 ±0.0050 Δ at pub 0.0240
23	GCN code ↗ GNN global	0.8260 ±0.0220	Scalable Expressiveness through Preprocessed Graph Perturbations code ↗
30	GIN GNN baseline actual #30 vs paper #111 81 places worse	0.8220 ±0.0210 best reported	actual #30 vs paper #111 81 places worse paper 0.8020 ±0.0190 best at pub: Scalable Expressiveness through Preprocessed Graph Perturbations (2024) Difference within reported uncertainty; not counted as underperformance. paper 0.8020 ±0.0190 Δ at pub 0.0200
33	DiffPool GNN baseline actual #33 vs paper #161 128 places worse	0.8213 ±0.0043 best reported	actual #33 vs paper #161 128 places worse paper 0.7890 ±0.0230 best at pub: Scalable Graph Compressed Convolutions (2024) Difference within reported uncertainty; not counted as underperformance. paper 0.7890 ±0.0230 Δ at pub 0.0323
68	GCMAE GNN baseline actual #68 vs paper #68	0.8132 ±0.0032 best reported	actual #68 vs paper #68 paper 0.8130 ±0.0030 best at pub: Generative and Contrastive Paradigms Are Complementary for Graph Self-Supervised Learning (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.8130 ±0.0030 Δ at pub 0.0002
75	GCC GNN baseline actual #75 vs paper #161 86 places worse	0.8110 best reported	actual #75 vs paper #161 86 places worse paper 0.7890 best at pub: Graph Self-Supervised Learning: A Survey (2021) paper 0.7890 Δ at pub 0.0220
80	GraphCL GNN baseline actual #80 vs paper #289 209 places worse	0.8108 ±0.0017 best reported	actual #80 vs paper #289 209 places worse paper 0.7140 ±0.0120 best at pub: Label-invariant Augmentation for Semi-Supervised Graph Classification (NeurIPS 2022) paper 0.7140 ±0.0120 Δ at pub 0.0968
103	AUG-MAE GNN baseline actual #103 vs paper #103	0.8050 ±0.0050	actual #103 vs paper #103 paper 0.8050 ±0.0050 prev best: Rethinking Graph Masked Autoencoders through Alignment and Uniformity (AAAI 2024) Improvement within reported uncertainty; not highlighted. paper 0.8050 ±0.0050 Δ at pub 0.0002
126	InfoGCL GNN baseline actual #126 vs paper #126	0.8001 ±0.0132 best reported	actual #126 vs paper #126 paper 0.8000 ±0.0130 best at pub: Self-Supervised Graph Learning via Spectral Bootstrapping and Laplacian-Based Augmentations (2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8000 ±0.0130 Δ at pub 0.0001
146	InfoGraph GNN baseline actual #146 vs paper #294 148 places worse	0.7940 best reported	actual #146 vs paper #294 148 places worse paper 0.7070 ±0.0110 best at pub: CogDL: A Comprehensive Library for Graph Deep Learning (WWW 2021) paper 0.7070 ±0.0110 Δ at pub 0.0870
151	MLP Trad global	0.7920 ±0.0240	Simplifying Graph Convolutional Networks with Redundancy-Free Neighbors
248	JOAO GNN baseline actual #248 vs paper #305 57 places worse	0.7535 ±0.0093 best reported	actual #248 vs paper #305 57 places worse paper 0.6950 ±0.0040 best at pub: Spectral Augmentation for Self-Supervised Learning on Graphs (ICLR 2022) paper 0.6950 ±0.0040 Δ at pub 0.0585

OGB

ogbn-products

Global leaderboard →

#	Model	Accuracy ↑	Paper
1	Jacobi GNN global	0.8968	Language Models as Messengers: Enhancing Message Passing in Heterophilic Graph Learning
2	GPR GNN global	0.8953	Language Models as Messengers: Enhancing Message Passing in Heterophilic Graph Learning
3	LEMP4HG ext data Hyb global	0.8939	Language Models as Messengers: Enhancing Message Passing in Heterophilic Graph Learning
4	SAGE GNN global	0.8933	Language Models as Messengers: Enhancing Message Passing in Heterophilic Graph Learning
30	GraphMAE GNN baseline actual #30 vs paper #194 164 places worse	0.8630 ±0.0007 best reported	actual #30 vs paper #194 164 places worse paper 0.7889 best at pub: Towards Graph Foundation Models: Learning Generalities Across Graphs via Task-Trees (ICML 2024) paper 0.7889 Δ at pub 0.0741
125	GraphMAE2 GNN baseline actual #125 vs paper #142 17 places worse	0.8108 ±0.0800 best reported	actual #125 vs paper #142 17 places worse paper 0.8009 best at pub: Quantizing Text-attributed Graphs for Semantic-Structural Integration (KDD 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8009 Δ at pub 0.0099
152	CORE Hyb ★ proposed	0.7982	This paper
351	MLP Trad global	0.6605 ±0.0020	Latent Graph Inference using Product Manifolds

SOTA performance (not significantly different from the best) Proposed method (this paper) Baseline / comparison (this paper) Global context (other papers) Best at pub — reported result worse than the best known when published (best reported in green) Best at pub — reported value improved over the best known when published

#	Model	Accuracy ↑	Paper
1	GraphSAGE+CNA code ↗ GNN global	0.9418 ±0.0033	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
2	GCN+CNA code ↗ GNN global	0.9366 ±0.0048	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
3	TransformerConv code ↗ GNN global	0.9350 ±0.0058	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
21	GCN GNN baseline actual #21 vs paper #573 552 places worse	0.9004 best reported	actual #21 vs paper #573 552 places worse paper 0.8150 ±0.0020 best at pub: Understanding Heterophily for Graph Neural Networks (ICML 2024) paper 0.8150 ±0.0020 Δ at pub 0.0854
39	SAGE GNN global	0.8911 ±0.0012	Efficient Text-Attributed Graph Learning through Selective Annotation and Graph Alignment
44	GAT GNN baseline actual #44 vs paper #424 380 places worse	0.8897 ±0.0014 best reported	actual #44 vs paper #424 380 places worse paper 0.8300 ±0.0070 best at pub: Adaptive Tokenization: On the Hop-Overpriority Problem in Tokenized Graph Learning Models (2025) paper 0.8300 ±0.0070 Δ at pub 0.0597
130	BGRL GNN baseline actual #130 vs paper #447 317 places worse	0.8730 ±0.0010 best reported	actual #130 vs paper #447 317 places worse paper 0.8270 ±0.0060 best at pub: Oversmoothing: A Nightmare for Graph Contrastive Learning? (2023) paper 0.8270 ±0.0060 Δ at pub 0.0460
134	GRACE GNN baseline actual #134 vs paper #536 402 places worse	0.8720 ±0.0020 best reported	actual #134 vs paper #536 402 places worse paper 0.8190 ±0.0040 best at pub: Oversmoothing: A Nightmare for Graph Contrastive Learning? (2023) paper 0.8190 ±0.0040 Δ at pub 0.0530
156	MVGRL GNN baseline actual #156 vs paper #365 209 places worse	0.8680 ±0.0050 best reported	actual #156 vs paper #365 209 places worse paper 0.8350 ±0.0040 best at pub: ENGAGE: Explanation Guided Data Augmentation for Graph Representation Learning (2023) best today: Adaptive Negative Scheduling for Graph Contrastive Learning (2026) paper 0.8350 ±0.0040 Δ at pub 0.0330 Δ today 0.0330
177	DGI GNN baseline actual #177 vs paper #486 309 places worse	0.8630 ±0.0020 best reported	actual #177 vs paper #486 309 places worse paper 0.8230 ±0.0060 best at pub: Oversmoothing: A Nightmare for Graph Contrastive Learning? (2023) paper 0.8230 ±0.0060 Δ at pub 0.0400
218	GMI GNN baseline actual #218 vs paper #424 206 places worse	0.8526 ±0.0008 best reported	actual #218 vs paper #424 206 places worse paper 0.8300 ±0.0020 best at pub: Node Representation Learning in Graph via Node-to-Neighbourhood Mutual Information Maximization (2022) paper 0.8300 ±0.0020 Δ at pub 0.0226
248	CCA-SSG GNN baseline actual #248 vs paper #303 55 places worse	0.8460 ±0.0070 best reported	actual #248 vs paper #303 55 places worse paper 0.8400 ±0.0040 best at pub: Oversmoothing: A Nightmare for Graph Contrastive Learning? (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.8400 ±0.0040 Δ at pub 0.0060
252	GraphMAE2 GNN baseline actual #252 vs paper #289 37 places worse	0.8450 ±0.0060 best reported	actual #252 vs paper #289 37 places worse paper 0.8410 ±0.0060 best at pub: Disentangled Generative Graph Representation Learning (2024) Difference within reported uncertainty; not counted as underperformance. paper 0.8410 ±0.0060 Δ at pub 0.0040
270	AUG-MAE GNN baseline	0.8430 ±0.0040	This paper
281	CORE Hyb ★ proposed	0.8420 ±0.0050	This paper
280	GraphMAE GNN baseline actual #280 vs paper #325 45 places worse	0.8420 ±0.0040 best reported	actual #280 vs paper #325 45 places worse paper 0.8370 ±0.0080 best at pub: Disentangled Generative Graph Representation Learning (2024) best today: Discrepancy-Aware Graph Mask Auto-Encoder (KDD 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.8370 ±0.0080 Δ at pub 0.0050 Δ today 0.0050
415	GCA GNN baseline actual #415 vs paper #550 135 places worse	0.8310 ±0.0030 best reported	actual #415 vs paper #550 135 places worse paper 0.8180 ±0.0020 best at pub: Graph Contrastive Learning with Implicit Augmentations (2022) paper 0.8180 ±0.0020 Δ at pub 0.0130
654	GAE GNN baseline actual #654 vs paper #810 156 places worse	0.8011 ±0.0022 best reported	actual #654 vs paper #810 156 places worse paper 0.7150 ±0.0040 best at pub: Graph Masked Autoencoders with Transformers (2022) paper 0.7150 ±0.0040 Δ at pub 0.0861
724	MLP Trad global	0.7768	Understanding Heterophily for Graph Neural Networks

#	Model	Accuracy ↑	Paper
1	GraphTARIF code ↗ Hyb global	0.9703 ±0.0019	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
2	GAT GNN baseline actual #2 vs paper #269 267 places worse	0.9669 ±0.0014 best reported	actual #2 vs paper #269 267 places worse paper 0.9260 ±0.0040 best at pub: SignGT: Signed Attention-based Graph Transformer for Graph Representation Learning (2023) best today: Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification (2026) paper 0.9260 ±0.0040 Δ at pub 0.0238 Δ today 0.0409
3	GAT+T code ↗ Hyb global	0.9660 ±0.0007	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
7	SAGE code ↗ GNN global	0.9643 ±0.0027	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
8	GCN GNN baseline actual #8 vs paper #294 286 places worse	0.9632 ±0.0008 best reported	actual #8 vs paper #294 286 places worse paper 0.9240 ±0.0020 best at pub: SignGT: Signed Attention-based Graph Transformer for Graph Representation Learning (2023) best today: Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification (2026) paper 0.9240 ±0.0020 Δ at pub 0.0236 Δ today 0.0392
109	CORE Hyb ★ proposed	0.9390 ±0.0040	This paper
149	ABGML GNN baseline actual #149 vs paper #149	0.9350 ±0.0040	actual #149 vs paper #149 paper 0.9350 ±0.0040 prev best: From Overfitting to Robustness: Quantity, Quality, and Variety Oriented Negative Sample Selection in Graph Contrastive Learning (2024) Improvement within reported uncertainty; not highlighted. paper 0.9350 ±0.0040 Δ at pub 0.0004
164	GraphMAE GNN baseline actual #164 vs paper #221 57 places worse	0.9341 ±0.0010 best reported	actual #164 vs paper #221 57 places worse paper 0.9300 ±0.0030 best at pub: Rethinking and Simplifying Bootstrapped Graph Latents (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.9300 ±0.0030 Δ at pub 0.0041
172	BGRL GNN baseline actual #172 vs paper #190 18 places worse	0.9337 ±0.0021 best reported	actual #172 vs paper #190 18 places worse paper 0.9320 ±0.0030 best at pub: Wiener Graph Deconvolutional Network Improves Graph Self-Supervised Learning (AAAI 2022) Difference within reported uncertainty; not counted as underperformance. paper 0.9320 ±0.0030 Δ at pub 0.0017
186	CCA-SSG GNN baseline actual #186 vs paper #205 19 places worse	0.9325 ±0.0021 best reported	actual #186 vs paper #205 19 places worse paper 0.9310 ±0.0010 best at pub: Wiener Graph Deconvolutional Network Improves Graph Self-Supervised Learning (AAAI 2022) Difference within reported uncertainty; not counted as underperformance. paper 0.9310 ±0.0010 Δ at pub 0.0015
191	GCA GNN baseline actual #191 vs paper #279 88 places worse	0.9320 ±0.0060 best reported	actual #191 vs paper #279 88 places worse paper 0.9250 ±0.0010 best at pub: Unifying Graph Contrastive Learning with Flexible Contextual Scopes (2022) best today: Decoupling Weighing and Selecting for Integrating Multiple Graph Pre-training Tasks (ICLR 2024) paper 0.9250 ±0.0010 Δ at pub 0.0070 Δ today 0.0070
192	GraphMAE2 GNN baseline	0.9320 ±0.0030	This paper
203	GRACE GNN baseline actual #203 vs paper #237 34 places worse	0.9313 ±0.0017 best reported	actual #203 vs paper #237 34 places worse paper 0.9280 ±0.0050 best at pub: Learn Beneficial Noise as Graph Augmentation (ICML 2025) Difference within reported uncertainty; not counted as underperformance. paper 0.9280 ±0.0050 Δ at pub 0.0033
206	MVGRL GNN baseline actual #206 vs paper #362 156 places worse	0.9310 ±0.0040 best reported	actual #206 vs paper #362 156 places worse paper 0.9170 ±0.0010 best at pub: Decoupling Weighing and Selecting for Integrating Multiple Graph Pre-training Tasks (ICLR 2024) paper 0.9170 ±0.0010 Δ at pub 0.0140
213	DGI GNN baseline actual #213 vs paper #366 153 places worse	0.9309 ±0.0008 best reported	actual #213 vs paper #366 153 places worse paper 0.9160 ±0.0020 best at pub: Noise-Resilient Unsupervised Graph Representation Learning via Multi-Hop Feature Quality Estimation (2024) paper 0.9160 ±0.0020 Δ at pub 0.0149
352	GAE GNN baseline actual #352 vs paper #366 14 places worse	0.9185 ±0.0006 best reported	actual #352 vs paper #366 14 places worse paper 0.9160 ±0.0010 best at pub: Graph Masked Autoencoders with Transformers (2022) paper 0.9160 ±0.0010 Δ at pub 0.0025
390	MLP code ↗ Trad global	0.9128 ±0.0063	Clarify Confused Nodes via Separated Learning code ↗
408	GMI GNN baseline actual #408 vs paper #416 8 places worse	0.9081 ±0.0015 best reported	actual #408 vs paper #416 8 places worse paper 0.9070 ±0.0020 best at pub: Localized Graph Contrastive Learning (2022) best today: Energy-Balanced Hyperspherical Graph Representation Learning via Structural Binding and Entropic Dispersion (2025) Difference within reported uncertainty; not counted as underperformance. paper 0.9070 ±0.0020 Δ at pub 0.0003 Δ today 0.0011