Geometric Evolution Graph Convolutional Networks: Enhancing Graph Representation Learning via Ricci Flow

Jicheng Ma, Yunyan Yang, Juan Zhao, Liang Zhao

Abstract

We introduce the Geometric Evolution Graph Convolutional Network (GEGCN), a novel framework that enhances graph representation learning through explicit modeling of geometric evolution on graph structures. Specifically, GEGCN leverages a Long Short-Term Memory (LSTM) network to capture the dynamic structural sequence generated by discrete Ricci flow, and infuses the learned dynamic representations into a graph convolutional network. Extensive experiments demonstrate that GEGCN achieves excellent performance on classification tasks across various benchmark datasets, including homophilic/heterophilic graphs, filtered graphs, and large-scale graphs.

Benchmark Results

Tested on 16 datasets 59 baseline values reported too low Non-standard metric on 2 datasets

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

#	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 ±0.0050 best at pub: Understanding Heterophily for Graph Neural Networks (ICML 2024) paper 0.8150 ±0.0050 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
55	JKNet GNN baseline actual #55 vs paper #325 270 places worse	0.8870 best reported	actual #55 vs paper #325 270 places worse paper 0.8370 ±0.0070 best at pub: Learned Low Precision Graph Neural Networks (2020) paper 0.8370 ±0.0070 Δ at pub 0.0500
77	GPRGNN GNN baseline actual #77 vs paper #315 238 places worse	0.8826 ±0.0050 best reported	actual #77 vs paper #315 238 places worse paper 0.8380 ±0.0090 best at pub: Polynomial Selection in Spectral Graph Neural Networks: An Error-Sum of Function Slices Approach (WWW 2024) best today: Partition-wise Graph Filtering: A Unified Perspective Through the Lens of Graph Coarsening (KDD 2025) paper 0.8380 ±0.0090 Δ at pub 0.0446 Δ today 0.0446
108	APPNP GNN baseline actual #108 vs paper #365 257 places worse	0.8787 best reported	actual #108 vs paper #365 257 places worse paper 0.8350 ±0.0070 best at pub: Orthogonal Graph Neural Networks (AAAI 2021) best today: Feature Transportation Improves Graph Neural Networks (AAAI 2023) paper 0.8350 ±0.0070 Δ at pub 0.0437 Δ today 0.0437
141	GraphSAGE GNN baseline actual #141 vs paper #688 547 places worse	0.8708 ±0.0085 best reported	actual #141 vs paper #688 547 places worse paper 0.7920 ±0.0770 best at pub: Large Language Model Meets Graph Neural Network in Knowledge Distillation (AAAI 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.7920 ±0.0770 Δ at pub 0.0788
163	GEGCN Hyb ★ proposed	0.8670 ±0.0120	This paper
266	RC-UFG (Hom) GNN baseline	0.8440 ±0.0070	This paper
350	UFGConv_R GNN baseline	0.8360 ±0.0060	This paper
428	UFGConv_S GNN baseline	0.8300 ±0.0050	This paper
458	CurvGN GNN baseline	0.8260 ±0.0060	This paper
508	MoNet GNN baseline actual #508 vs paper #558 50 places worse	0.8220 ±0.0070 best reported	actual #508 vs paper #558 50 places worse paper 0.8170 ±0.0040 best at pub: GRAND: Graph Neural Diffusion (ICML 2021) Difference within reported uncertainty; not counted as underperformance. paper 0.8170 ±0.0040 Δ at pub 0.0050
724	MLP Trad baseline actual #724 vs paper #890 166 places worse	0.7768 best reported	actual #724 vs paper #890 166 places worse paper 0.5510 ±0.0140 best at pub: Understanding Heterophily for Graph Neural Networks (ICML 2024) paper 0.5510 ±0.0140 Δ at pub 0.2258

#	Model	Accuracy ↑	Paper
1	CNA code ↗ GNN global	0.9575 ±0.0058	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
2	IS-GIB code ↗ Hyb global	0.9390 ±0.0187	Individual and Structural Graph Information Bottlenecks for Out-of-Distribution Generalization code ↗
3	EERM code ↗ GNN global	0.9112 ±0.0145	Individual and Structural Graph Information Bottlenecks for Out-of-Distribution Generalization code ↗
20	APPNP GNN baseline actual #20 vs paper #140 120 places worse	0.8142 ±0.0026 best reported	actual #20 vs paper #140 120 places worse paper 0.7590 ±0.0060 best at pub: Feature Expansion for Graph Neural Networks (ICML 2023) paper 0.7590 ±0.0060 Δ at pub 0.0552
32	GPRGNN GNN baseline actual #32 vs paper #140 108 places worse	0.8102 ±0.0071 best reported	actual #32 vs paper #140 108 places worse paper 0.7590 ±0.0070 best at pub: NTFormer: A Composite Node Tokenized Graph Transformer for Node Classification (2024) paper 0.7590 ±0.0070 Δ at pub 0.0512
35	GAT GNN baseline actual #35 vs paper #388 353 places worse	0.8080 ±0.0026 best reported	actual #35 vs paper #388 353 places worse paper 0.7250 ±0.0070 best at pub: Feature Expansion for Graph Neural Networks (ICML 2023) paper 0.7250 ±0.0070 Δ at pub 0.0830
44	GCN GNN baseline actual #44 vs paper #561 517 places worse	0.8033 ±0.0088 best reported	actual #44 vs paper #561 517 places worse paper 0.7090 ±0.0050 best at pub: NTFormer: A Composite Node Tokenized Graph Transformer for Node Classification (2024) paper 0.7090 ±0.0050 Δ at pub 0.0943
61	GraphSAGE GNN baseline actual #61 vs paper #496 435 places worse	0.7923 ±0.0053 best reported	actual #61 vs paper #496 435 places worse paper 0.7160 ±0.0190 best at pub: Gophormer: Ego-Graph Transformer for Node Classification (2021) paper 0.7160 ±0.0190 Δ at pub 0.0763
62	JKNet GNN baseline actual #62 vs paper #388 326 places worse	0.7920 best reported	actual #62 vs paper #388 326 places worse paper 0.7250 ±0.0040 best at pub: Tackling Over-Smoothing for General Graph Convolutional Networks (2020) paper 0.7250 ±0.0040 Δ at pub 0.0670
121	GEGCN Hyb ★ proposed	0.7660 ±0.0110	This paper
259	SAGE code ↗ GNN global	0.7361 ±0.0190	Efficient Tuning and Inference for Large Language Models on Textual Graphs code ↗
350	MLP Trad baseline actual #350 vs paper #834 484 places worse	0.7289 ±0.0010 best reported	actual #350 vs paper #834 484 places worse paper 0.5910 ±0.0120 best at pub: Gumbel-MPNN: Graph Rewiring with Gumbel-Softmax (2025) paper 0.5910 ±0.0120 Δ at pub 0.1379
365	UFGConv_R GNN baseline	0.7270 ±0.0060	This paper
397	RC-UFG (Hom) GNN baseline	0.7250 ±0.0070	This paper
506	CurvGN GNN baseline	0.7150 ±0.0080	This paper
540	MoNet GNN baseline	0.7120 ±0.0070	This paper
556	UFGConv_S GNN baseline	0.7100 ±0.0060	This paper

#	Model	Accuracy ↑	Paper
1	LGTL GT global	0.9518 ±0.0011	Adaptive Tokenization: On the Hop-Overpriority Problem in Tokenized Graph Learning Models
2	MixHop (LO) code ↗ GNN global	0.9513 ±0.0038	LOGIN: A Large Language Model Consulted Graph Neural Network Training Framework code ↗
3	GraphSAGE (LO) code ↗ GNN global	0.9511 ±0.0036	LOGIN: A Large Language Model Consulted Graph Neural Network Training Framework code ↗
17	APPNP GNN baseline actual #17 vs paper #428 411 places worse	0.9355 ±0.0060 best reported	actual #17 vs paper #428 411 places worse paper 0.8020 ±0.0030 best at pub: LOGIN: A Large Language Model Consulted Graph Neural Network Training Framework (2024) paper 0.8020 ±0.0030 Δ at pub 0.1335
18	GCN GNN baseline actual #18 vs paper #585 567 places worse	0.9296 ±0.0015 best reported	actual #18 vs paper #585 567 places worse paper 0.7900 ±0.0030 best at pub: Adaptive Tokenization: On the Hop-Overpriority Problem in Tokenized Graph Learning Models (2025) paper 0.7900 ±0.0030 Δ at pub 0.1396
20	GAT GNN baseline actual #20 vs paper #585 565 places worse	0.9233 ±0.0018 best reported	actual #20 vs paper #585 565 places worse paper 0.7900 ±0.0030 best at pub: Adaptive Tokenization: On the Hop-Overpriority Problem in Tokenized Graph Learning Models (2025) paper 0.7900 ±0.0030 Δ at pub 0.1333
24	JKNet GNN baseline actual #24 vs paper #313 289 places worse	0.9160 best reported	actual #24 vs paper #313 289 places worse paper 0.8260 ±0.0050 best at pub: Tackling Over-Smoothing for General Graph Convolutional Networks (2020) paper 0.8260 ±0.0050 Δ at pub 0.0900
101	GPRGNN GNN baseline actual #101 vs paper #321 220 places worse	0.8911 ±0.0008 best reported	actual #101 vs paper #321 220 places worse paper 0.8230 ±0.0020 best at pub: Feature Expansion for Graph Neural Networks (ICML 2023) paper 0.8230 ±0.0020 Δ at pub 0.0681
116	SAGE GNN global	0.8881 ±0.0015	Efficient Text-Attributed Graph Learning through Selective Annotation and Graph Alignment
125	GraphSAGE GNN baseline actual #125 vs paper #716 591 places worse	0.8864 ±0.0030 best reported	actual #125 vs paper #716 591 places worse paper 0.7740 ±0.0220 best at pub: Non-backtracking Graph Neural Networks (TMLR 2023) paper 0.7740 ±0.0220 Δ at pub 0.1124
189	GEGCN Hyb ★ proposed	0.8740 ±0.0040	This paper
210	MLP Trad baseline actual #210 vs paper #827 617 places worse	0.8690 ±0.0074 best reported	actual #210 vs paper #827 617 places worse paper 0.7140 ±0.0080 best at pub: On the Equivalence of Graph Convolution and Mixup (TMLR 2023) paper 0.7140 ±0.0080 Δ at pub 0.1550
305	RC-UFG (Hom) GNN baseline	0.8290 ±0.0020	This paper
446	MoNet GNN baseline actual #446 vs paper #628 182 places worse	0.8006 ±0.0055 best reported	actual #446 vs paper #628 182 places worse paper 0.7860 ±0.0050 best at pub: Not All Neighbors Are Worth Attending to: Graph Selective Attention Networks for Semi-supervised Learning (2022) paper 0.7860 ±0.0050 Δ at pub 0.0146
488	UFGConv_R GNN baseline	0.7960 ±0.0040	This paper
525	UFGConv_S GNN baseline	0.7940 ±0.0040	This paper
616	CurvGN GNN baseline	0.7880 ±0.0060	This paper

#	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 #508 506 places worse	0.9669 ±0.0014 best reported	actual #2 vs paper #508 506 places worse paper 0.8510 ±0.0230 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 0.8510 ±0.0230 Δ at pub 0.0988 Δ today 0.1159
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 #395 387 places worse	0.9632 ±0.0008 best reported	actual #8 vs paper #395 387 places worse paper 0.9120 ±0.0120 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 0.9120 ±0.0120 Δ at pub 0.0356 Δ today 0.0512
21	APPNP GNN baseline actual #21 vs paper #497 476 places worse	0.9573 ±0.0012 best reported	actual #21 vs paper #497 476 places worse paper 0.8700 ±0.0090 best at pub: NTFormer: A Composite Node Tokenized Graph Transformer for Node Classification (2024) paper 0.8700 ±0.0090 Δ at pub 0.0873
33	GPRGNN GNN baseline actual #33 vs paper #497 464 places worse	0.9549 ±0.0014 best reported	actual #33 vs paper #497 464 places worse paper 0.8700 ±0.0090 best at pub: SignGT: Signed Attention-based Graph Transformer for Graph Representation Learning (2023) paper 0.8700 ±0.0090 Δ at pub 0.0849
56	GraphSAGE GNN baseline actual #56 vs paper #378 322 places worse	0.9508 ±0.0026 best reported	actual #56 vs paper #378 322 places worse paper 0.9140 ±0.0140 best at pub: A Quasi-Wasserstein Loss for Learning Graph Neural Networks (WWW 2023) paper 0.9140 ±0.0140 Δ at pub 0.0368
93	GEGCN Hyb ★ proposed	0.9410 ±0.0050	This paper
152	RC-UFG (Hom) GNN baseline	0.9350 ±0.0070	This paper
280	CurvGN GNN baseline	0.9250 ±0.0050	This paper
282	UFGConv_R GNN baseline	0.9250 ±0.0020	This paper
331	UFGConv_S GNN baseline	0.9210 ±0.0050	This paper
364	JKNet GNN baseline actual #364 vs paper #501 137 places worse	0.9169 best reported	actual #364 vs paper #501 137 places worse paper 0.8610 ±0.0110 best at pub: Structure-Aware DropEdge Towards Deep Graph Convolutional Networks (2023) paper 0.8610 ±0.0110 Δ at pub 0.0559
390	MLP Trad baseline actual #390 vs paper #558 168 places worse	0.9128 ±0.0063 best reported	actual #390 vs paper #558 168 places worse paper 0.6960 ±0.0380 best at pub: Clarify Confused Nodes via Separated Learning (2023) paper 0.6960 ±0.0380 Δ at pub 0.2168
393	MoNet GNN baseline actual #393 vs paper #395 2 places worse	0.9124 ±0.0130 best reported	actual #393 vs paper #395 2 places worse paper 0.9120 ±0.0130 best at pub: Building Shortcuts between Distant Nodes with Biaffine Mapping for Graph Convolutional Networks (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.9120 ±0.0130 Δ at pub 0.0004

#	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 GNN baseline actual #6 vs paper #139 133 places worse	0.9593 ±0.0025 best reported	actual #6 vs paper #139 133 places worse paper 0.9130 ±0.0280 best at pub: Graph Wave Networks (WWW 2025) paper 0.9130 ±0.0280 Δ at pub 0.0463
12	GCN GNN baseline actual #12 vs paper #148 136 places worse	0.9514 ±0.0025 best reported	actual #12 vs paper #148 136 places worse paper 0.9110 ±0.0050 best at pub: Graph Wave Networks (WWW 2025) paper 0.9110 ±0.0050 Δ at pub 0.0404
21	GAT GNN baseline actual #21 vs paper #156 135 places worse	0.9437 ±0.0028 best reported	actual #21 vs paper #156 135 places worse paper 0.9050 ±0.0060 best at pub: Graph Wave Networks (WWW 2025) paper 0.9050 ±0.0060 Δ at pub 0.0387
23	RC-UFG (Hom) GNN baseline	0.9420 ±0.0090	This paper
25	APPNP GNN baseline actual #25 vs paper #136 111 places worse	0.9400 ±0.0030 best reported	actual #25 vs paper #136 111 places worse paper 0.9150 ±0.0010 best at pub: Diffusion on Graph: Augmentation of Graph Structure for Node Classification (TMLR 2025) paper 0.9150 ±0.0010 Δ at pub 0.0250
60	GEGCN Hyb ★ proposed	0.9320 ±0.0040	This paper
70	UFGConv_R GNN baseline	0.9300 ±0.0070	This paper
86	CurvGN GNN baseline	0.9290 ±0.0040	This paper
88	MLP Trad baseline actual #88 vs paper #171 83 places worse	0.9280 ±0.0020 best reported	actual #88 vs paper #171 83 places worse paper 0.8830 ±0.0070 best at pub: Revisiting the Necessity of Graph Learning and Common Graph Benchmarks (2024) paper 0.8830 ±0.0070 Δ at pub 0.0450
121	UFGConv_S GNN baseline	0.9210 ±0.0020	This paper
124	GPRGNN GNN baseline actual #124 vs paper #131 7 places worse	0.9203 ±0.0061 best reported	actual #124 vs paper #131 7 places worse paper 0.9180 ±0.0010 best at pub: Towards Understanding Graph Neural Networks: An Algorithm Unrolling Perspective (2022) Difference within reported uncertainty; not counted as underperformance. paper 0.9180 ±0.0010 Δ at pub 0.0023
138	GIN GNN global	0.9140 ±0.0020	Learning Graph Augmentations to Learn Graph Representations
150	JKNet GNN baseline actual #150 vs paper #150	0.9110 ±0.0030	actual #150 vs paper #150 paper 0.9110 ±0.0030 prev best: Towards Understanding Graph Neural Networks: An Algorithm Unrolling Perspective (2022) Improvement within reported uncertainty; not highlighted. paper 0.9110 ±0.0030 Δ at pub 0.0012
155	MoNet GNN baseline actual #155 vs paper #155	0.9082 ±0.0160 best reported	actual #155 vs paper #155 paper 0.9080 ±0.0060 best at pub: Building Shortcuts between Distant Nodes with Biaffine Mapping for Graph Convolutional Networks (2023) Difference within reported uncertainty; not counted as underperformance. paper 0.9080 ±0.0060 Δ at pub 0.0002

#	Model	Accuracy ↑	Paper
1	GIANT-XRT code ↗ Hyb global	0.7612 ±0.0016	Node Feature Extraction by Self-Supervised Multi-scale Neighborhood Prediction code ↗
2	GEGCN Hyb ★ proposed	0.7576 ±0.0064	This paper
3	AGDN GNN baseline	0.7522 ±0.0007	This paper
5	UniMP GT baseline	0.7450 ±0.0005	This paper
15	GAT GNN baseline actual #15 vs paper #19 4 places worse	0.7365 ±0.0011 best reported	actual #15 vs paper #19 4 places worse paper 0.7330 ±0.0016 best at pub: Fully-inductive Node Classification on Arbitrary Graphs (ICLR 2024) paper 0.7330 ±0.0016 Δ at pub 0.0035
23	GCN GNN baseline actual #23 vs paper #23	0.7300 ±0.0017	actual #23 vs paper #23 paper 0.7300 ±0.0017 prev best: Rethinking and Simplifying Bootstrapped Graph Latents (2023) paper 0.7300 ±0.0017 Δ at pub 0.0126
25	GraphSAGE GNN baseline actual #25 vs paper #25	0.7277 ±0.0016	actual #25 vs paper #25 paper 0.7277 ±0.0016 prev best: All Against Some: Efficient Integration of Large Language Models for Message Passing in Graph Neural Networks (2024) paper 0.7277 ±0.0016 Δ at pub 0.0122
26	SAGE code ↗ GNN global	0.7267 ±0.0031	Node Identifiers: Compact, Discrete Representations for Efficient Graph Learning code ↗
68	Node2Vec Walk baseline actual #68 vs paper #68	0.7129 ±0.0013	actual #68 vs paper #68 paper 0.7129 ±0.0013 prev best: Graph Barlow Twins: A self-supervised representation learning framework for graphs (2021) paper 0.7129 ±0.0013 Δ at pub 0.0122
107	MLP Trad baseline actual #107 vs paper #107	0.5765 ±0.0012	actual #107 vs paper #107 paper 0.5765 ±0.0012 prev best: Rethinking and Simplifying Bootstrapped Graph Latents (2023) paper 0.5765 ±0.0012 Δ at pub 0.0135

#	Model	Accuracy ↑	Paper
1	P^2GNN Hyb global	0.9672 ±0.0147	$P^2$GNN: Two Prototype Sets to boost GNN Performance
2	DoG code ↗ ext data GNN global	0.9660 ±0.0210	Diffusion on Graph: Augmentation of Graph Structure for Node Classification code ↗
3	GPF-plus code ↗ ext data GNN global	0.9583 ±0.0419	ProG: A Graph Prompt Learning Benchmark code ↗
34	MLP Trad global	0.9230 ±0.0070	Design your own universe: a physics-informed agnostic method for enhancing graph neural networks
53	GraphSAGE code ↗ GNN global	0.9000 ±0.0365	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
179	GNRF GNN baseline	0.8333 ±0.0608	This paper
257	GCN GNN baseline actual #257 vs paper #489 232 places worse	0.7933 ±0.0447 best reported	actual #257 vs paper #489 232 places worse paper 0.6119 ±0.0049 best at pub: Graph Wave Networks (WWW 2025) paper 0.6119 ±0.0049 Δ at pub 0.1814
352	+FA GNN baseline actual #352 vs paper #443 91 places worse	0.7300 ±0.0330 best reported	actual #352 vs paper #443 91 places worse paper 0.6482 ±0.0029 best at pub: Graph Representation Learning via Aggregation Enhancement (2022) paper 0.6482 ±0.0029 Δ at pub 0.0818
387	GEGCN Hyb ★ proposed	0.7027 ±0.0069	This paper
422	SDRF + Undirected GNN baseline	0.6702 ±0.0040	This paper
445	SDRF Trad baseline actual #445 vs paper #445	0.6446 ±0.0038	actual #445 vs paper #445 paper 0.6446 ±0.0038 prev best: Local Virtual Nodes for Alleviating Over-Squashing in Graph Neural Networks (2025) paper 0.6446 ±0.0038 Δ at pub 0.1295
457	DIGL + Undirected GNN baseline	0.6354 ±0.0038	This paper
458	Undirected GNN baseline	0.6338 ±0.0087	This paper
478	DIGL Trad baseline actual #478 vs paper #478	0.6203 ±0.0043	actual #478 vs paper #478 paper 0.6203 ±0.0043 prev best: Local Virtual Nodes for Alleviating Over-Squashing in Graph Neural Networks (2025) paper 0.6203 ±0.0043 Δ at pub 0.1046

#	Model	Accuracy ↑	Paper
1	P^2GNN Hyb global	0.9541 ±0.0272	$P^2$GNN: Two Prototype Sets to boost GNN Performance
2	ACM-GCN code ↗ GNN global	0.9480 ±0.0380	Heterophily-informed Message Passing code ↗
3	RF-GCN GNN global	0.9429	Simplifying Graph Convolutional Networks with Redundancy-Free Neighbors
24	MLP Trad global	0.9136 ±0.0070	Node-oriented Spectral Filtering for Graph Neural Networks
138	SAGE GNN global	0.8269	Language Models as Messengers: Enhancing Message Passing in Heterophilic Graph Learning
170	GCN GNN baseline actual #170 vs paper #459 289 places worse	0.8082 ±0.0360 best reported	actual #170 vs paper #459 289 places worse paper 0.5269 ±0.0021 best at pub: Bayesian Neighborhood Adaptation for Graph Neural Networks (TMLR 2025) paper 0.5269 ±0.0021 Δ at pub 0.2813
254	GNRF GNN baseline	0.7305 ±0.0352	This paper
295	GEGCN Hyb ★ proposed	0.6861 ±0.0026	This paper
378	DIGL + Undirected GNN baseline	0.5954 ±0.0064	This paper
399	+FA GNN baseline	0.5829 ±0.0049	This paper
401	DIGL Trad baseline actual #401 vs paper #401	0.5826 ±0.0050	actual #401 vs paper #401 paper 0.5826 ±0.0050 prev best: Local Virtual Nodes for Alleviating Over-Squashing in Graph Neural Networks (2025) paper 0.5826 ±0.0050 Δ at pub 0.0961
410	SDRF + Undirected GNN baseline	0.5754 ±0.0034	This paper
431	SDRF Trad baseline actual #431 vs paper #431	0.5460 ±0.0039	actual #431 vs paper #431 paper 0.5460 ±0.0039 prev best: Local Virtual Nodes for Alleviating Over-Squashing in Graph Neural Networks (2025) paper 0.5460 ±0.0039 Δ at pub 0.0698
452	Undirected GNN baseline	0.5320 ±0.0053	This paper

#	Model	Accuracy ↑	Paper
1	DiGGR Hyb global	0.4535 ±0.0353	Disentangled Generative Graph Representation Learning
2	TRIGON code ↗ Hyb global	0.4402 ±0.0028	Dynamic Triangulation-Based Graph Rewiring for Graph Neural Networks code ↗
3	DoG code ↗ ext data GNN global	0.4320 ±0.0120	Diffusion on Graph: Augmentation of Graph Structure for Node Classification code ↗
42	MLP Trad global	0.4026 ±0.0108	Heterophily-Aware Graph Attention Network
68	GraphSAGE code ↗ GNN global	0.3899 ±0.0085	A Quasi-Wasserstein Loss for Learning Graph Neural Networks code ↗
126	GEGCN Hyb ★ proposed	0.3718 ±0.0030	This paper
135	GCN GNN baseline actual #135 vs paper #443 308 places worse	0.3690 best reported	actual #135 vs paper #443 308 places worse paper 0.2384 ±0.0043 best at pub: Beyond Low-frequency Information in Graph Convolutional Networks (AAAI 2021) paper 0.2384 ±0.0043 Δ at pub 0.1306
210	GNRF GNN baseline	0.3553 ±0.0068	This paper
320	DIGL Trad baseline actual #320 vs paper #436 116 places worse	0.3122 ±0.0047 best reported	actual #320 vs paper #436 116 places worse paper 0.2477 ±0.0032 best at pub: Dynamic Triangulation-Based Graph Rewiring for Graph Neural Networks (2025) paper 0.2477 ±0.0032 Δ at pub 0.0645
387	SDRF Trad baseline	0.2842 ±0.0075	This paper
388	SDRF + Undirected GNN baseline	0.2835 ±0.0006	This paper
427	DIGL + Undirected GNN baseline	0.2545 ±0.0030	This paper
442	+FA GNN baseline	0.2414 ±0.0043	This paper
476	Undirected GNN baseline	0.2145 ±0.0047	This paper

#	Model	Accuracy ↑	Paper
1	Hetero-S (GAT 8-layer) code ↗ Hyb global	0.8693	The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs code ↗
2	Dir-GNN+CNA code ↗ GNN global	0.8586 ±0.0180	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
3	GAT code ↗ GNN global	0.8560	The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs code ↗
52	GCN GNN baseline actual #52 vs paper #248 196 places worse	0.7033 best reported	actual #52 vs paper #248 196 places worse paper 0.4133 ±0.0018 best at pub: Understanding Heterophily for Graph Neural Networks (ICML 2024) paper 0.4133 ±0.0018 Δ at pub 0.2900
108	SAGE code ↗ GNN global	0.6340 ±0.0300	Scale-aware Message Passing For Graph Node Classification code ↗
116	DIGL Trad baseline actual #116 vs paper #244 128 places worse	0.6208 ±0.0054 best reported	actual #116 vs paper #244 128 places worse paper 0.4202 ±0.0013 best at pub: Local Virtual Nodes for Alleviating Over-Squashing in Graph Neural Networks (2025) paper 0.4202 ±0.0013 Δ at pub 0.2006
122	SDRF Trad baseline actual #122 vs paper #240 118 places worse	0.6119 ±0.0055 best reported	actual #122 vs paper #240 118 places worse paper 0.4273 ±0.0015 best at pub: Local Virtual Nodes for Alleviating Over-Squashing in Graph Neural Networks (2025) paper 0.4273 ±0.0015 Δ at pub 0.1846
128	GEGCN Hyb ★ proposed	0.6072 ±0.0016	This paper
169	MLP Trad global	0.5429	Understanding Heterophily for Graph Neural Networks
172	GNRF GNN baseline	0.5382 ±0.0302	This paper
229	SDRF + Undirected GNN baseline	0.4446 ±0.0017	This paper
241	DIGL + Undirected GNN baseline	0.4268 ±0.0012	This paper
242	+FA GNN baseline	0.4267 ±0.0017	This paper
245	Undirected GNN baseline	0.4202 ±0.0030	This paper

#	Model	Accuracy ↑	Paper
1	IBG-NN GNN global	0.7763 ±0.0179	Efficient Learning on Large Graphs using a Densifying Regularity Lemma
2	Dir-GNN+CNA code ↗ GNN global	0.7747 ±0.0128	Graph Neural Networks Need Cluster-Normalize-Activate Modules code ↗
3	DHGNN GNN global	0.7684 ±0.0161	Directed Homophily-Aware Graph Neural Network
10	GPR-GNN GNN baseline actual #10 vs paper #217 207 places worse	0.7480 ±0.0050 best reported	actual #10 vs paper #217 207 places worse paper 0.3511 ±0.0124 best at pub: Efficient Learning on Large Graphs using a Densifying Regularity Lemma (2025) paper 0.3511 ±0.0124 Δ at pub 0.3969
11	FSGNN GNN baseline actual #11 vs paper #205 194 places worse	0.7416 best reported	actual #11 vs paper #205 194 places worse paper 0.3592 ±0.0132 best at pub: Simplifying approach to Node Classification in Graph Neural Networks (2021) paper 0.3592 ±0.0132 Δ at pub 0.3824
19	GloGNN GNN baseline actual #19 vs paper #212 193 places worse	0.7121 ±0.0184 best reported	actual #19 vs paper #212 193 places worse paper 0.3551 ±0.0165 best at pub: Directed Homophily-Aware Graph Neural Network (2025) paper 0.3551 ±0.0165 Δ at pub 0.3570
39	H2GCN GNN baseline actual #39 vs paper #217 178 places worse	0.6190 ±0.0140 best reported	actual #39 vs paper #217 178 places worse paper 0.3510 ±0.0115 best at pub: Efficient Learning on Large Graphs using a Densifying Regularity Lemma (2025) paper 0.3510 ±0.0115 Δ at pub 0.2680
46	GCN GNN global	0.6067	Understanding Heterophily for Graph Neural Networks
67	GAT GNN baseline actual #67 vs paper #211 144 places worse	0.5630 ±0.0220 best reported	actual #67 vs paper #211 144 places worse paper 0.3562 ±0.0206 best at pub: Restructuring Graphs for Higher Homophily via Adaptive Spectral Clustering (AAAI 2022) paper 0.3562 ±0.0206 Δ at pub 0.2068
139	FAGCN GNN baseline actual #139 vs paper #173 34 places worse	0.4259 ±0.0069 best reported	actual #139 vs paper #173 34 places worse paper 0.3895 ±0.0199 best at pub: Gradient Gating for Deep Multi-Rate Learning on Graphs (ICLR 2022) best today: Graph Neural Reaction Diffusion Models Submitted to the editors June 2023. Accepted in March 2024. (2024) paper 0.3895 ±0.0199 Δ at pub 0.0364 Δ today 0.0364
155	GBK-GNN GNN baseline	0.4108 ±0.0227	This paper
165	GEGCN Hyb ★ proposed	0.4003 ±0.0171	This paper
172	MLP code ↗ Trad global	0.3930 ±0.0079	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
201	ResNet Trad baseline	0.3655 ±0.0182	This paper
236	GNRF GNN baseline	0.3135 ±0.0179	This paper

#	Model	Accuracy ↑	Paper
1	DMD++ Hyb global	0.9190 ±0.0260	When Graph Neural Networks Meet Dynamic Mode Decomposition
2	F-GREAD-BS code ↗ GNN global	0.9059 ±0.0380	Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND code ↗
3	APPNP GNN global	0.9010 ±0.0350	When Graph Neural Networks Meet Dynamic Mode Decomposition
18	GNRF GNN baseline	0.8820 ±0.0384	This paper
37	MLP code ↗ Trad global	0.8629 ±0.0450	Is Homophily a Necessity for Graph Neural Networks? code ↗
63	GCN GNN baseline actual #63 vs paper #188 125 places worse	0.8310 ±0.0320 best reported	actual #63 vs paper #188 125 places worse paper 0.5460 ±0.0086 best at pub: Restructuring Graphs for Higher Homophily via Adaptive Spectral Clustering (AAAI 2022) paper 0.5460 ±0.0086 Δ at pub 0.2850
69	GraphSAGE code ↗ GNN global	0.8118 ±0.0556	Graph Navier–Stokes Networks code ↗
140	GEGCN Hyb ★ proposed	0.6739 ±0.0093	This paper
181	SDRF + Undirected GNN baseline	0.5655 ±0.0086	This paper
184	SDRF Trad baseline	0.5551 ±0.0027	This paper
185	+FA GNN baseline	0.5548 ±0.0062	This paper
202	DIGL + Undirected GNN baseline	0.5223 ±0.0054	This paper
206	Undirected GNN baseline	0.5137 ±0.0115	This paper
213	DIGL Trad baseline	0.4953 ±0.0027	This paper

#	Model	Accuracy ↑	Paper
1	Dir-Poly code ↗ GNN global	0.9451 ±0.0022	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
2	D-CDE code ↗ GNN global	0.9387 ±0.0041	Distributed-Order Fractional Graph Operating Network code ↗
3	LargeScaleNet code ↗ Hyb global	0.9358 ±0.0024	Scale-aware Message Passing For Graph Node Classification code ↗
26	GraphSAGE code ↗ GNN global	0.9106 ±0.0027	Directed Semi-Simplicial Learning with Applications to Brain Activity Decoding code ↗
57	GEGCN Hyb ★ proposed	0.8674 ±0.0102	This paper
83	GAT GNN baseline actual #83 vs paper #84 1 place worse	0.8145 ±0.0009 best reported	actual #83 vs paper #84 1 place worse paper 0.8087 ±0.0030 best at pub: Gumbel-MPNN: Graph Rewiring with Gumbel-Softmax (2025) paper 0.8087 ±0.0030 Δ at pub 0.0058
92	FSGNN GNN baseline	0.7992 ±0.0056	This paper
94	GNRF GNN baseline	0.7972 ±0.0077	This paper
107	GBK-GNN GNN baseline actual #107 vs paper #109 2 places worse	0.7587 ±0.0043 best reported	actual #107 vs paper #109 2 places worse paper 0.7457 ±0.0047 best at pub: Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024) best today: Distributed-Order Fractional Graph Operating Network (NeurIPS 2024) paper 0.7457 ±0.0047 Δ at pub 0.0130 Δ today 0.0130
108	FAGCN GNN baseline actual #108 vs paper #142 34 places worse	0.7571 ±0.0139 best reported	actual #108 vs paper #142 34 places worse paper 0.6522 ±0.0056 best at pub: NTFormer: A Composite Node Tokenized Graph Transformer for Node Classification (2024) paper 0.6522 ±0.0056 Δ at pub 0.1049
114	GPR-GNN GNN baseline actual #114 vs paper #144 30 places worse	0.7337 ±0.0068 best reported	actual #114 vs paper #144 30 places worse paper 0.6485 ±0.0027 best at pub: Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024) best today: Distributed-Order Fractional Graph Operating Network (NeurIPS 2024) paper 0.6485 ±0.0027 Δ at pub 0.0852 Δ today 0.0852
129	H2GCN GNN baseline actual #129 vs paper #164 35 places worse	0.6809 ±0.0029 best reported	actual #129 vs paper #164 35 places worse paper 0.6011 ±0.0052 best at pub: Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024) best today: Distributed-Order Fractional Graph Operating Network (NeurIPS 2024) paper 0.6011 ±0.0052 Δ at pub 0.0798 Δ today 0.0798
135	MLP code ↗ Trad global	0.6665 ±0.0047	HyperAggregation: Aggregating over Graph Edges with Hypernetworks code ↗
141	ResNet Trad baseline	0.6588 ±0.0038	This paper
154	GloGNN GNN baseline actual #154 vs paper #167 13 places worse	0.6385 ±0.0049 best reported	actual #154 vs paper #167 13 places worse paper 0.5963 ±0.0069 best at pub: Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024) best today: Distributed-Order Fractional Graph Operating Network (NeurIPS 2024) paper 0.5963 ±0.0069 Δ at pub 0.0422 Δ today 0.0422

#	Model	Accuracy ↑	Paper
1	UniGAP global	0.6375 ±0.0110	UniGAP: A Universal and Adaptive Graph Upsampling Approach to Mitigate Over-Smoothing in Node …
2	H2GCN + HalfHop GNN global	0.5961 ±0.0100	UniGAP: A Universal and Adaptive Graph Upsampling Approach to Mitigate Over-Smoothing in Node …
3	GraphSAGE + HalfHop GNN global	0.5897 ±0.0150	UniGAP: A Universal and Adaptive Graph Upsampling Approach to Mitigate Over-Smoothing in Node …
7	H2GCN GNN baseline actual #7 vs paper #196 189 places worse	0.5681 ±0.0090 best reported	actual #7 vs paper #196 189 places worse paper 0.3647 ±0.0023 best at pub: UniGAP: A Universal and Adaptive Graph Upsampling Approach to Mitigate Over-Smoothing in Node Classification Tasks (2024) paper 0.3647 ±0.0023 Δ at pub 0.2034
12	SAGE code ↗ GNN global	0.5569 ±0.0023	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
14	GAT GNN baseline actual #14 vs paper #105 91 places worse	0.5559 ±0.0027 best reported	actual #14 vs paper #105 91 places worse paper 0.4909 ±0.0063 best at pub: Relieving the Over-Aggregating Effect in Graph Transformers (NeurIPS 2025) best today: Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification (2026) paper 0.4909 ±0.0063 Δ at pub 0.0361 Δ today 0.0650
48	GEGCN Hyb ★ proposed	0.5277 ±0.0073	This paper
50	FSGNN GNN baseline	0.5274 ±0.0083	This paper
84	MLP Trad global	0.5058	Understanding Heterophily for Graph Neural Networks
103	GPR-GNN GNN baseline actual #103 vs paper #148 45 places worse	0.4937 ±0.0071 best reported	actual #103 vs paper #148 45 places worse paper 0.4488 ±0.0034 best at pub: Piecewise Constant Spectral Graph Neural Network (TMLR 2025) paper 0.4488 ±0.0034 Δ at pub 0.0449
137	FAGCN GNN baseline actual #137 vs paper #156 19 places worse	0.4632 ±0.0250 best reported	actual #137 vs paper #156 19 places worse paper 0.4412 ±0.0030 best at pub: Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024) best today: Distributed-Order Fractional Graph Operating Network (NeurIPS 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.4412 ±0.0030 Δ at pub 0.0220 Δ today 0.0220
140	GBK-GNN GNN baseline	0.4598 ±0.0071	This paper
141	ResNet Trad baseline	0.4590 ±0.0052	This paper
187	GNRF GNN baseline	0.4004 ±0.0095	This paper
196	GloGNN GNN baseline actual #196 vs paper #196	0.3728 ±0.0066 best reported	actual #196 vs paper #196 paper 0.3689 ±0.0014 best at pub: Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024) best today: Distributed-Order Fractional Graph Operating Network (NeurIPS 2024) Difference within reported uncertainty; not counted as underperformance. paper 0.3689 ±0.0014 Δ at pub 0.0039 Δ today 0.0039

#	Model	Accuracy ↑	Paper
1	GraphTARIF code ↗ Hyb global	0.9903 ±0.0019	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
2	BuNN code ↗ Hyb global	0.9899 ±0.0016	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
3	M3Dphormer code ↗ GT global	0.9827 ±0.0020	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
5	GAT GNN baseline actual #5 vs paper #18 13 places worse	0.9786 ±0.0037 best reported	actual #5 vs paper #18 13 places worse paper 0.9201 ±0.0068 best at pub: Multi-Scale Harmonic Encoding for Feature-Wise Graph Message Passing (2025) best today: Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification (2026) paper 0.9201 ±0.0068 Δ today 0.0585
9	GCN code ↗ GNN global	0.9744 ±0.0017	Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification code ↗
18	GEGCN Hyb ★ proposed	0.9225 ±0.0093	This paper
22	GBK-GNN GNN baseline	0.9085 ±0.0058	This paper
24	FSGNN GNN baseline	0.9008 ±0.0070	This paper
25	H2GCN GNN baseline	0.8971 ±0.0031	This paper
26	FAGCN GNN baseline	0.8817 ±0.0073	This paper
30	GPR-GNN GNN baseline	0.8624 ±0.0061	This paper
33	GNRF GNN baseline	0.8481 ±0.0125	This paper
43	MLP code ↗ Trad global	0.8000	Learning Posterior Predictive Distributions for Node Classification from Synthetic Graph Priors code ↗
58	GloGNN GNN baseline	0.5108 ±0.0123	This paper
60	ResNet Trad baseline	0.5089 ±0.0139	This paper

#	Model	Accuracy ↑	Paper
1	MLP code ↗ Trad global	0.9733 ±0.0006	Learning Posterior Predictive Distributions for Node Classification from Synthetic Graph Priors code ↗
2	AGS-GNN GNN global ✎	0.9727	Graph Neural Networks for Graphs with Heterophily: A Survey
3	MeanGNN code ↗ GNN global	0.9716 ±0.0006	Equivariance Everywhere All At Once: A Recipe for Graph Foundation Models code ↗
4	GCN code ↗ GNN global	0.9715 ±0.0004	Learning Posterior Predictive Distributions for Node Classification from Synthetic Graph Priors code ↗
5	GAT GNN baseline actual #5 vs paper #31 26 places worse	0.9713 ±0.0005 best reported	actual #5 vs paper #31 26 places worse paper 0.7743 ±0.0120 best at pub: Equivariance Everywhere All At Once: A Recipe for Graph Foundation Models (2025) paper 0.7743 ±0.0120 Δ at pub 0.1970
29	FSGNN GNN baseline	0.7886 ±0.0092	This paper
31	GEGCN Hyb ★ proposed	0.7773 ±0.0041	This paper
32	FAGCN GNN baseline	0.7724 ±0.0126	This paper
37	GBK-GNN GNN baseline	0.7447 ±0.0086	This paper
45	GNRF GNN baseline	0.7038 ±0.0375	This paper
46	ResNet Trad baseline	0.7034 ±0.0076	This paper
50	GloGNN GNN baseline	0.6574 ±0.0119	This paper
51	H2GCN GNN baseline	0.6359 ±0.0146	This paper
52	GPR-GNN GNN baseline	0.5548 ±0.0091	This paper

Abstract

Benchmark Results

Classic

OGB

Heterophilic Graphs

Heterophily Benchmark