HeroFilter: Adaptive Spectral Graph Filter for Varying Heterophilic Relations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Shuaicheng, Wang, Haohui, Lin, Junhong, Guo, Xiaojie, Zhu, Yada, Zhang, Si, Fu, Dongqi, Zhou, Dawei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915549009674240
author Zhang, Shuaicheng
Wang, Haohui
Lin, Junhong
Guo, Xiaojie
Zhu, Yada
Zhang, Si
Fu, Dongqi
Zhou, Dawei
author_facet Zhang, Shuaicheng
Wang, Haohui
Lin, Junhong
Guo, Xiaojie
Zhu, Yada
Zhang, Si
Fu, Dongqi
Zhou, Dawei
contents Graph heterophily, where connected nodes have different labels, has attracted significant interest recently. Most existing works adopt a simplified approach - using low-pass filters for homophilic graphs and high-pass filters for heterophilic graphs. However, we discover that the relationship between graph heterophily and spectral filters is more complex - the optimal filter response varies across frequency components and does not follow a strict monotonic correlation with heterophily degree. This finding challenges conventional fixed filter designs and suggests the need for adaptive filtering to preserve expressiveness in graph embeddings. Formally, natural questions arise: Given a heterophilic graph G, how and to what extent will the varying heterophily degree of G affect the performance of GNNs? How can we design adaptive filters to fit those varying heterophilic connections? Our theoretical analysis reveals that the average frequency response of GNNs and graph heterophily degree do not follow a strict monotonic correlation, necessitating adaptive graph filters to guarantee good generalization performance. Hence, we propose [METHOD NAME], a simple yet powerful GNN, which extracts information across the heterophily spectrum and combines salient representations through adaptive mixing. [METHOD NAME]'s superior performance achieves up to 9.2% accuracy improvement over leading baselines across homophilic and heterophilic graphs.
format Preprint
id arxiv_https___arxiv_org_abs_2510_10864
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HeroFilter: Adaptive Spectral Graph Filter for Varying Heterophilic Relations
Zhang, Shuaicheng
Wang, Haohui
Lin, Junhong
Guo, Xiaojie
Zhu, Yada
Zhang, Si
Fu, Dongqi
Zhou, Dawei
Machine Learning
Artificial Intelligence
Social and Information Networks
Graph heterophily, where connected nodes have different labels, has attracted significant interest recently. Most existing works adopt a simplified approach - using low-pass filters for homophilic graphs and high-pass filters for heterophilic graphs. However, we discover that the relationship between graph heterophily and spectral filters is more complex - the optimal filter response varies across frequency components and does not follow a strict monotonic correlation with heterophily degree. This finding challenges conventional fixed filter designs and suggests the need for adaptive filtering to preserve expressiveness in graph embeddings. Formally, natural questions arise: Given a heterophilic graph G, how and to what extent will the varying heterophily degree of G affect the performance of GNNs? How can we design adaptive filters to fit those varying heterophilic connections? Our theoretical analysis reveals that the average frequency response of GNNs and graph heterophily degree do not follow a strict monotonic correlation, necessitating adaptive graph filters to guarantee good generalization performance. Hence, we propose [METHOD NAME], a simple yet powerful GNN, which extracts information across the heterophily spectrum and combines salient representations through adaptive mixing. [METHOD NAME]'s superior performance achieves up to 9.2% accuracy improvement over leading baselines across homophilic and heterophilic graphs.
title HeroFilter: Adaptive Spectral Graph Filter for Varying Heterophilic Relations
topic Machine Learning
Artificial Intelligence
Social and Information Networks
url https://arxiv.org/abs/2510.10864