Order-Robust Class Incremental Learning: Graph-Driven Dynamic Similarity Grouping

TL;DR

This paper introduces GDDSG, a graph-based method that dynamically groups classes by similarity to improve robustness and accuracy in class incremental learning, especially under varying class orderings.

Contribution

The paper presents a novel graph-driven grouping approach that enhances class incremental learning by reducing order sensitivity and improving model robustness.

Findings

GDDSG significantly reduces order sensitivity in CIL.

The method improves model accuracy and anti-forgetting capabilities.

Experimental results outperform existing CIL methods.

Abstract

Class Incremental Learning (CIL) aims to enable models to learn new classes sequentially while retaining knowledge of previous ones. Although current methods have alleviated catastrophic forgetting (CF), recent studies highlight that the performance of CIL models is highly sensitive to the order of class arrival, particularly when sequentially introduced classes exhibit high inter-class similarity. To address this critical yet understudied challenge of class order sensitivity, we first extend existing CIL frameworks through theoretical analysis, proving that grouping classes with lower pairwise similarity during incremental phases significantly improves model robustness to order variations. Building on this insight, we propose Graph-Driven Dynamic Similarity Grouping (GDDSG), a novel method that employs graph coloring algorithms to dynamically partition classes into similarity-constrained groups. Each group trains an isolated CIL sub-model and constructs meta-features for class group identification. Experimental results demonstrate that our method effectively addresses the issue of class order sensitivity while achieving optimal performance in both model accuracy and anti-forgetting capability. Our code is available at https://github.com/AIGNLAI/GDDSG.