Robust Graph Condensation via Classification Complexity Mitigation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Luo, Jiayi, Sun, Qingyun, Yang, Beining, Yuan, Haonan, Fu, Xingcheng, Ma, Yanbiao, Li, Jianxin, Yu, Philip S.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908669267935232
author Luo, Jiayi
Sun, Qingyun
Yang, Beining
Yuan, Haonan
Fu, Xingcheng
Ma, Yanbiao
Li, Jianxin
Yu, Philip S.
author_facet Luo, Jiayi
Sun, Qingyun
Yang, Beining
Yuan, Haonan
Fu, Xingcheng
Ma, Yanbiao
Li, Jianxin
Yu, Philip S.
contents Graph condensation (GC) has gained significant attention for its ability to synthesize smaller yet informative graphs. However, existing studies often overlook the robustness of GC in scenarios where the original graph is corrupted. In such cases, we observe that the performance of GC deteriorates significantly, while existing robust graph learning technologies offer only limited effectiveness. Through both empirical investigation and theoretical analysis, we reveal that GC is inherently an intrinsic-dimension-reducing process, synthesizing a condensed graph with lower classification complexity. Although this property is critical for effective GC performance, it remains highly vulnerable to adversarial perturbations. To tackle this vulnerability and improve GC robustness, we adopt the geometry perspective of graph data manifold and propose a novel Manifold-constrained Robust Graph Condensation framework named MRGC. Specifically, we introduce three graph data manifold learning modules that guide the condensed graph to lie within a smooth, low-dimensional manifold with minimal class ambiguity, thereby preserving the classification complexity reduction capability of GC and ensuring robust performance under universal adversarial attacks. Extensive experiments demonstrate the robustness of \ModelName\ across diverse attack scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2510_26451
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Robust Graph Condensation via Classification Complexity Mitigation
Luo, Jiayi
Sun, Qingyun
Yang, Beining
Yuan, Haonan
Fu, Xingcheng
Ma, Yanbiao
Li, Jianxin
Yu, Philip S.
Machine Learning
Artificial Intelligence
Graph condensation (GC) has gained significant attention for its ability to synthesize smaller yet informative graphs. However, existing studies often overlook the robustness of GC in scenarios where the original graph is corrupted. In such cases, we observe that the performance of GC deteriorates significantly, while existing robust graph learning technologies offer only limited effectiveness. Through both empirical investigation and theoretical analysis, we reveal that GC is inherently an intrinsic-dimension-reducing process, synthesizing a condensed graph with lower classification complexity. Although this property is critical for effective GC performance, it remains highly vulnerable to adversarial perturbations. To tackle this vulnerability and improve GC robustness, we adopt the geometry perspective of graph data manifold and propose a novel Manifold-constrained Robust Graph Condensation framework named MRGC. Specifically, we introduce three graph data manifold learning modules that guide the condensed graph to lie within a smooth, low-dimensional manifold with minimal class ambiguity, thereby preserving the classification complexity reduction capability of GC and ensuring robust performance under universal adversarial attacks. Extensive experiments demonstrate the robustness of \ModelName\ across diverse attack scenarios.
title Robust Graph Condensation via Classification Complexity Mitigation
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2510.26451