The reviewed record of science sign in
Pith

arxiv: 2109.05641 · v1 · pith:2HK4RES6 · submitted 2021-09-12 · cs.LG · cs.SI

Is Heterophily A Real Nightmare For Graph Neural Networks To Do Node Classification?

Sitao Luan , Chenqing Hua , Qincheng Lu , Jiaqi Zhu , Mingde Zhao , Shuyuan Zhang , Xiao-Wen Chang , Doina Precup This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:2HK4RES6record.jsonopen to challenge →

classification cs.LG cs.SI
keywords gnnsheterophilygraphmetricsharmfulnetworksneuraltasks
0
0 comments X
read the original abstract

Graph Neural Networks (GNNs) extend basic Neural Networks (NNs) by using the graph structures based on the relational inductive bias (homophily assumption). Though GNNs are believed to outperform NNs in real-world tasks, performance advantages of GNNs over graph-agnostic NNs seem not generally satisfactory. Heterophily has been considered as a main cause and numerous works have been put forward to address it. In this paper, we first show that not all cases of heterophily are harmful for GNNs with aggregation operation. Then, we propose new metrics based on a similarity matrix which considers the influence of both graph structure and input features on GNNs. The metrics demonstrate advantages over the commonly used homophily metrics by tests on synthetic graphs. From the metrics and the observations, we find some cases of harmful heterophily can be addressed by diversification operation. With this fact and knowledge of filterbanks, we propose the Adaptive Channel Mixing (ACM) framework to adaptively exploit aggregation, diversification and identity channels in each GNN layer to address harmful heterophily. We validate the ACM-augmented baselines with 10 real-world node classification tasks. They consistently achieve significant performance gain and exceed the state-of-the-art GNNs on most of the tasks without incurring significant computational burden.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Evaluating LLMs on Large-Scale Graph Property Estimation via Random Walks

    cs.LG 2026-05 unverdicted novelty 7.0

    EstGraph benchmark evaluates LLMs on estimating properties of very large graphs from random-walk samples that fit in context limits.

  2. Enhancing LLMs for Graph Tasks via Graph-aware LoRA Generation

    cs.LG 2026-06 unverdicted novelty 6.0

    GaRA generates task-specific LoRA weight updates conditioned on graph structures to enable better whole-graph encoding in LLMs for zero-shot graph learning.

  3. SIGMA: An Efficient Heterophilous Graph Neural Network with Fast Global Aggregation

    cs.LG 2023-05 unverdicted novelty 6.0

    SIGMA integrates SimRank for one-time global similarity aggregation in heterophilous GNNs, achieving O(n) complexity and reported 5x speedup on large graphs with SOTA accuracy.

  4. Attention-based graph neural networks: a survey

    cs.SI 2026-05 unverdicted novelty 5.0

    The survey groups attention-based GNNs into three stages—graph recurrent attention networks, graph attention networks, and graph transformers—while reviewing architectures and future directions.

  5. Graph Neural Networks for Graphs with Heterophily: A Survey

    cs.LG 2022-02 unverdicted novelty 5.0

    A survey proposing a systematic taxonomy of GNNs for heterophilic graphs along with analyses of their relations to other graph research domains.

  6. Graph Rewiring in GNNs to Mitigate Over-Squashing and Over-Smoothing: A Survey

    cs.LG 2024-11 unverdicted novelty 2.0

    A survey compiling graph rewiring techniques for mitigating over-squashing and over-smoothing in GNNs.