mlsanityData to 7 Oct 2026

The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs

arXiv 2406.12539 ↗First posted 18 Jun 2024cs.LG cs.AIcited by 6

Not on the map. Why?

Abstract (arXiv)

Graph Neural Networks (GNNs) have become pivotal tools for a range of graph-based learning tasks. Notably, most current GNN architectures operate under the assumption of homophily, whether explicitly or implicitly. While this underlying assumption is frequently adopted, it is not universally applicable, which can result in potential shortcomings in learning effectiveness. In this paper, for the first time, we transfer the prevailing concept of ``one node one receptive field" to the heterophilic graph. By constructing a proxy label predictor, we enable each node to possess a latent prediction distribution, which assists connected nodes in determining whether they should aggregate their associated neighbors. Ultimately, every node can have its own unique aggregation hop and pattern, much like each snowflake is unique and possesses its own characteristics. Based on observations, we innovatively introduce the Heterophily Snowflake Hypothesis and provide an effective solution to guide and facilitate research on heterophilic graphs and beyond. We conduct comprehensive experiments including (1) main results on 10 graphs with varying heterophily ratios across 10 backbones; (2) scalability on various deep GNN backbones (SGC, JKNet, etc.) across various large number of layers (2,4,6,8,16,32 layers); (3) comparison with conventional snowflake hypothesis; (4) efficiency comparison with existing graph pruning algorithms. Our observations show that our framework acts as a versatile operator for diverse tasks. It can be integrated into various GNN frameworks, boosting performance in-depth and offering an explainable approach to choosing the optimal network depth. The source code is available at urlhttps://github.com/bingreeky/HeteroSnoH.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 55 of its 75 references that are in the map. First posted on arXiv 18 Jun 2024.

Builds on

55 of 75 references.

References · 55

  1. 2017Graph Attention Networkscited by 8,854
  2. 2017Inductive Representation Learning on Large Graphscited by 6,284
  3. 2018How Powerful are Graph Neural Networks?cited by 4,355
  4. 2019A Comprehensive Survey on Graph Neural Networkscited by 2,906
  5. 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,597
  6. 2018Graph Neural Networks: A Review of Methods and Applicationscited by 1,571
  7. 2019Simplifying Graph Convolutional Networkscited by 1,470
  8. 2018Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learningcited by 1,099
  9. 2018Pitfalls of Graph Neural Network Evaluationcited by 994
  10. 2018Predict then Propagate: Graph Neural Networks meet Personalized PageRankcited by 987
  11. 2018Representation Learning on Graphs with Jumping Knowledge Networkscited by 875
  12. 2020Simple and Deep Graph Convolutional Networkscited by 857
38 more
  1. 2018Link Prediction Based on Graph Neural Networkscited by 829
  2. 2020Geom-GCN: Geometric Graph Convolutional Networkscited by 821
  3. 2018SNIP: Single-shot Network Pruning based on Connection Sensitivitycited by 802
  4. 2019GNNExplainer: Generating Explanations for Graph Neural Networkscited by 735
  5. 2019Heterogeneous Graph Attention Networkcited by 732
  6. 2019Graph Neural Networks for Social Recommendationcited by 711
  7. 2020Adaptive Universal Generalized PageRank Graph Neural Networkcited by 592
  8. 2018FastGCN: Fast Learning with Graph Convolutional Networks via Importance Samplingcited by 583
  9. 2019Measuring and Relieving the Over-smoothing Problem for Graph Neural Networks from the Topological Viewcited by 544
  10. 2019Multi-scale Attributed Node Embeddingcited by 543
  11. 2020Graph Neural Networks in Recommender Systems: A Surveycited by 493
  12. 2019Graph U-Netscited by 489
  13. 2019MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixingcited by 442
  14. 2019Self-Attention Graph Poolingcited by 421
  15. 2018Session-based Recommendation with Graph Neural Networkscited by 312
  16. 2021Large Scale Learning on Non-Homophilous Graphs: New Benchmarks and Strong Simple Methodscited by 309
  17. 2020Towards Deeper Graph Neural Networkscited by 296
  18. 2021Graph Neural Network for Traffic Forecasting: A Surveycited by 260
  19. 2019Position-aware Graph Neural Networkscited by 203
  20. 2020Graph Neural Networks with Heterophilycited by 202
  21. 2022Finding Global Homophily in Graph Neural Networks When Meeting Heterophilycited by 179
  22. 2022Graph Neural Networks for Graphs with Heterophily: A Surveycited by 175
  23. 2019Understanding Attention and Generalization in Graph Neural Networkscited by 169
  24. 2019Graph Neural Tangent Kernel: Fusing Graph Neural Networks with Graph Kernelscited by 160
  25. 2019ASAP: Adaptive Structure Aware Pooling for Learning Hierarchical Graph Representationscited by 156
  26. 2021A Unified Lottery Ticket Hypothesis for Graph Neural Networkscited by 118
  27. 2018Attention-based Graph Neural Network for Semi-supervised Learningcited by 100
  28. 2020Building powerful and equivariant graph neural networks with structural message-passingcited by 95
  29. 2021Breaking the Limit of Graph Neural Networks by Improving the Assortativity of Graphs with Local Mixing Patternscited by 95
  30. 2020Non-Local Graph Neural Networkscited by 91
  31. 2021New Benchmarks for Learning on Non-Homophilous Graphscited by 81
  32. 2021SimPLE: Similar Pseudo Label Exploitation for Semi-Supervised Classificationcited by 68
  33. 2018N-GCN: Multi-scale Graph Convolution for Semi-supervised Node Classificationcited by 54
  34. 2021A Survey of Machine Learning for Computer Architecture and Systemscited by 53
  35. 2021Tree Decomposed Graph Neural Networkcited by 35
  36. 2021Graph Pointer Neural Networkscited by 27
  37. 2020Representative Graph Neural Networkcited by 19
  38. 2021Label-Wise Graph Convolutional Network for Heterophilic Graphscited by 17

50 most cited of 55 All 55 (JSON)

Built on by

Cited by year

Citing papers per year: 2024: 3; 2025: 3.

6 papers.

  1. 2024Rethinking Fair Graph Neural Networks from Re-balancingcited by 11
  2. 2025Message-Passing State-Space Models: Improving Graph Learning with Modern Sequence Modelingcited by 9
  3. 2025Beyond Fixed Depth: Adaptive Graph Neural Networks for Node Classification Under Varying Homophilycited by 5
  4. 2024LAMP: Learnable Meta-Path Guided Adversarial Contrastive Learning for Heterogeneous Graphscited by 0
  5. 2024Fast Track to Winning Tickets: Repowering One-Shot Pruning for Graph Neural Networkscited by 0
  6. 2025TANGO: Graph Neural Dynamics via Learned Energy and Tangential Flowscited by 0