Efficient and Robust Continual Graph Learning for Graph Classification in Biology

TL;DR

This paper introduces PSCGL, a novel continual graph learning framework that improves biological graph classification by enhancing robustness, efficiency, and security against backdoor attacks through perturbation and sparsification strategies.

Contribution

The paper presents PSCGL, a new continual learning method for biological graphs that incorporates perturbation and sparsification to improve performance and security.

Findings

PSCGL effectively retains knowledge across multiple biological graph classification tasks.

The framework reduces storage needs while maintaining high classification accuracy.

PSCGL inherently defends against graph backdoor attacks.

Abstract

Graph classification is essential for understanding complex biological systems, where molecular structures and interactions are naturally represented as graphs. Traditional graph neural networks (GNNs) perform well on static tasks but struggle in dynamic settings due to catastrophic forgetting. We present Perturbed and Sparsified Continual Graph Learning (PSCGL), a robust and efficient continual graph learning framework for graph data classification, specifically targeting biological datasets. We introduce a perturbed sampling strategy to identify critical data points that contribute to model learning and a motif-based graph sparsification technique to reduce storage needs while maintaining performance. Additionally, our PSCGL framework inherently defends against graph backdoor attacks, which is crucial for applications in sensitive biological contexts. Extensive experiments on biological datasets demonstrate that PSCGL not only retains knowledge across tasks but also enhances the efficiency and robustness of graph classification models in biology.