Adversarial-Robustness-Guided Graph Pruning

TL;DR

This paper introduces a scalable graph pruning method guided by adversarial robustness to improve the resilience and efficiency of spectral clustering by removing vulnerable edges.

Contribution

It presents a novel adversarial-robustness-guided graph pruning framework that enhances graph learning by explicitly pruning edges susceptible to adversarial attacks.

Findings

Improves spectral clustering robustness against adversarial perturbations.

Enhances computational efficiency of graph learning methods.

Produces sparser graphs that maintain or improve clustering quality.

Abstract

Graph learning plays a central role in many data mining and machine learning tasks, such as manifold learning, data representation and analysis, dimensionality reduction, clustering, and visualization. In this work, we propose a highly scalable, adversarial-robustness-guided graph pruning framework for learning graph topologies from data. By performing a spectral adversarial robustness evaluation, our method aims to learn sparse, undirected graphs that help the underlying algorithms resist noise and adversarial perturbations. In particular, we explicitly identify and prune edges that are most vulnerable to adversarial attacks. We use spectral clustering, one of the most representative graph-based machine learning algorithms, to evaluate the proposed framework. Compared with prior state-of-the-art graph learning approaches, the proposed method is more scalable and significantly improves both the computational efficiency and the solution quality of spectral clustering.