| HoloNets: Spectral Convolutions do extend to Directed Graphs | ✓ Link | 76.71±1.92 | FaberNet | 2023-10-03 |
| Improving Graph Neural Networks by Learning Continuous Edge Directions | ✓ Link | 75.32±1.82 | CoED | 2024-10-18 |
| Edge Directionality Improves Learning on Heterophilic Graphs | ✓ Link | 75.31±1.92 | Dir-GNN | 2023-05-17 |
| Simple Truncated SVD based Model for Node Classification on Heterophilic Graphs | | 74.17±1.83 | HLP Concat | 2021-06-24 |
| Improving Graph Neural Networks with Simple Architecture Design | ✓ Link | 74.10±1.89 | FSGNN (8-hop) | 2021-05-17 |
| Diffusion-Jump GNNs: Homophiliation via Learnable Metric Filters | ✓ Link | 73.48±1.59 | DJ-GNN | 2023-06-29 |
| Make Heterophily Graphs Better Fit GNN: A Graph Rewiring Approach | | 72.24±1.52 | H2GCN+DHGR | 2022-09-17 |
| Beyond Homophily with Graph Echo State Networks | | 71.2±1.5 | Graph ESN | 2022-10-27 |
| Self-attention Dual Embedding for Graphs with Heterophily | | 68.20±1.57 | SADE-GCN | 2023-05-28 |
| Universal Deep GNNs: Rethinking Residual Connection in GNNs from a Path Decomposition Perspective for Preventing the Over-smoothing | | 68.13±2.59 | UDGNN (GCN) | 2022-05-30 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 67.4 ± 2.21 | ACMII-GCN++ | 2022-10-14 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 67.07 ± 1.65 | ACMII-GCN+ | 2022-10-14 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 67.06 ± 1.66 | ACM-GCN++ | 2022-10-14 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 66.98 ± 1.71 | ACM-GCN+ | 2022-10-14 |
| Transitivity-Preserving Graph Representation Learning for Bridging Local Connectivity and Role-based Similarity | ✓ Link | 66.96 ±2.49 | UGT | 2023-08-18 |
| Graph Neural Reaction Diffusion Models | | 65.62 ± 2.33 | RDGNN-I | 2024-06-16 |
| SignGT: Signed Attention-based Graph Transformer for Graph Representation Learning | | 64.25±1.48 | SignGT | 2023-10-17 |
| CN-Motifs Perceptive Graph Neural Networks | | 63.60±1.96 | CNMPGNN | 2021-11-15 |
| Sign is Not a Remedy: Multiset-to-Multiset Message Passing for Learning on Heterophilic Graphs | ✓ Link | 63.60 ± 1.7 | M2M-GNN | 2024-05-31 |
| Label-Wise Graph Convolutional Network for Heterophilic Graphs | ✓ Link | 62.6±1.6 | LW-GCN | 2021-10-15 |
| Ordered GNN: Ordering Message Passing to Deal with Heterophily and Over-smoothing | ✓ Link | 62.44±1.96 | Ordered GNN | 2023-02-03 |
| Heterophilous Distribution Propagation for Graph Neural Networks | | 62.07 ± 1.57 | HDP | 2024-05-31 |
| GCNH: A Simple Method For Representation Learning On Heterophilous Graphs | ✓ Link | 61.85±1.54 | GCNH | 2023-04-21 |
| Large Scale Learning on Non-Homophilous Graphs: New Benchmarks and Strong Simple Methods | ✓ Link | 61.81 ± 1.80 | LINKX | 2021-10-27 |
| Refining Latent Homophilic Structures over Heterophilic Graphs for Robust Graph Convolution Networks | | 60.27±1.2 | LHS | 2023-12-27 |
| CAT: A Causally Graph Attention Network for Trimming Heterophilic Graph | ✓ Link | 59.3±1.8 | CATv3-sup | 2023-12-14 |
| Non-Local Graph Neural Networks | ✓ Link | 59.0 ± 1.2 | NLGCN | 2020-05-29 |
| Finding Global Homophily in Graph Neural Networks When Meeting Heterophily | ✓ Link | 57.88±1.76– | GloGNN++ | 2022-05-15 |
| The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs | ✓ Link | 57.83 | JKNet + Hetero-S (8 layers) | 2024-06-18 |
| Finding Global Homophily in Graph Neural Networks When Meeting Heterophily | ✓ Link | 57.54±1.39 | GloGNN | 2022-05-15 |
| Enhancing Intra-class Information Extraction for Heterophilous Graphs: One Neural Architecture Search Approach | | 57.32±1.89 | IIE-GNN | 2022-11-20 |
| Non-Local Graph Neural Networks | ✓ Link | 56.8 ± 2.5 | NLGAT | 2020-05-29 |
| Neural Sheaf Diffusion: A Topological Perspective on Heterophily and Oversmoothing in GNNs | ✓ Link | 56.34 ± 1.32 | O(d)-NSD | 2022-02-09 |
| Restructuring Graph for Higher Homophily via Adaptive Spectral Clustering | | 56.3 ± 2.2 | LCS-GAT | 2022-06-06 |
| GraphRARE: Reinforcement Learning Enhanced Graph Neural Network with Relative Entropy | | 55.90±1.39 | GCN-RARE (λ=1.0) | 2023-12-15 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 55.19 ± 1.49 | ACM-GCN | 2022-10-14 |
| Two Sides of the Same Coin: Heterophily and Oversmoothing in Graph Convolutional Neural Networks | ✓ Link | 55.17 ± 1.58 | GGCN | 2021-02-12 |
| Transfer Entropy in Graph Convolutional Neural Networks | ✓ Link | 55.04±1.64 | TE-GCNN | 2024-06-08 |
| Neural Sheaf Diffusion: A Topological Perspective on Heterophily and Oversmoothing in GNNs | ✓ Link | 54.78 ± 1.81 | Diag-NSD | 2022-02-09 |
| Learn from Heterophily: Heterophilous Information-enhanced Graph Neural Network | ✓ Link | 54.78 ± 1.58 | HiGNN | 2024-03-26 |
| Neural Sheaf Diffusion: A Topological Perspective on Heterophily and Oversmoothing in GNNs | ✓ Link | 53.17 ± 1.31 | Gen-NSD | 2022-02-09 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 51.8 ± 1.5 | ACMII-GCN | 2022-10-14 |
| Breaking the Limit of Graph Neural Networks by Improving the Assortativity of Graphs with Local Mixing Patterns | ✓ Link | 48.85 ± 0.78 | WRGAT | 2021-06-11 |
| Adaptive Universal Generalized PageRank Graph Neural Network | ✓ Link | 46.31 ± 2.46 | GPRGCN | 2020-06-14 |
| Breaking the Entanglement of Homophily and Heterophily in Semi-supervised Node Classification | | 45.2±1.3 | ADPA | 2023-12-07 |
| Sheaf Neural Networks with Connection Laplacians | ✓ Link | 45.19±1.57 | Conn-NSD | 2022-06-17 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 45.00 ± 1.4 | ACM-SGC-1 | 2022-10-14 |
| MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing | ✓ Link | 43.80 ± 1.48 | MixHop | 2019-04-30 |
| Revisiting Heterophily For Graph Neural Networks | ✓ Link | 40.02 ± 0.96 | ACM-SGC-2 | 2022-10-14 |
| Heterophilic Graph Neural Networks Optimization with Causal Message-passing | | 39.78±0.91 | Gprompt+CausalMP | 2024-11-21 |
| Simple and Deep Graph Convolutional Networks | ✓ Link | 38.47 ± 1.58 | GCNII | 2020-07-04 |
| Geom-GCN: Geometric Graph Convolutional Networks | ✓ Link | 38.14 | Geom-GCN-P | 2020-02-13 |
| Understanding over-squashing and bottlenecks on graphs via curvature | ✓ Link | 37.05±0.17 | SDRF | 2021-11-29 |
| Geom-GCN: Geometric Graph Convolutional Networks | ✓ Link | 36.24 | Geom-GCN-S | 2020-02-13 |
| Non-Local Graph Neural Networks | ✓ Link | 33.7 ± 1.5 | NLMLP | 2020-05-29 |
| Geom-GCN: Geometric Graph Convolutional Networks | ✓ Link | 33.32 | Geom-GCN-I | 2020-02-13 |
| Beyond Homophily in Graph Neural Networks: Current Limitations and Effective Designs | ✓ Link | 32.33 ± 1.94 | H2GCN-2 | 2020-06-20 |
| Beyond Low-frequency Information in Graph Convolutional Networks | ✓ Link | 30.83 ± 0.69 | FAGCN | 2021-01-04 |
| Beyond Homophily in Graph Neural Networks: Current Limitations and Effective Designs | ✓ Link | 28.98 ± 1.97 | H2GCN-1 | 2020-06-20 |