Breaking Forgetting: Training-Free Few-Shot Class-Incremental Learning via Conditional Diffusion

TL;DR

This paper introduces a training-free FSCIL framework using conditional diffusion instead of gradient updates, significantly reducing computational costs and improving adaptation to new classes with minimal data.

Contribution

It proposes a novel diffusion-based approach for FSCIL that eliminates the need for gradient optimization, addressing catastrophic forgetting and training cost issues.

Findings

Achieves state-of-the-art results on FSCIL benchmarks.

Reduces computational and memory overhead.

Effectively mitigates catastrophic forgetting.

Abstract

Efforts to overcome catastrophic forgetting in Few-Shot Class-Incremental Learning (FSCIL) have primarily focused on developing more effective gradient-based optimization strategies. In contrast, little attention has been paid to the training cost explosion that inevitably arises as the number of novel classes increases, a consequence of relying on gradient learning even under extreme data scarcity. More critically, since FSCIL typically provides only a few samples for each new class, gradient-based updates not only induce severe catastrophic forgetting on base classes but also hinder adaptation to novel ones. This paper seeks to break this long-standing limitation by asking: Can we design a training-free FSCIL paradigm that entirely removes gradient optimization? We provide an affirmative answer by uncovering an intriguing connection between gradient-based optimization and the Conditional Diffusion process. Building on this observation, we propose a Conditional Diffusion-driven FSCIL (CD-FSCIL) framework that substitutes the conventional gradient update process with a diffusion-based generative transition, enabling training-free incremental adaptation while effectively mitigating forgetting. Furthermore, to enhance representation under few-shot constraints, we introduce a multimodal learning strategy that integrates visual features with natural language descriptions automatically generated by Large Language Models (LLMs). This synergy substantially alleviates the sample scarcity issue and improves generalization across novel classes. Extensive experiments on mainstream FSCIL benchmarks demonstrate that our method not only achieves state-of-the-art performance but also drastically reduces computational and memory overhead, marking a paradigm shift toward training-free continual adaptation.