MOS: Model Surgery for Pre-Trained Model-Based Class-Incremental Learning

TL;DR

This paper introduces MOS, a model surgery approach that uses task-specific adapters and a retrieval mechanism to prevent catastrophic forgetting in pre-trained models during class-incremental learning, achieving state-of-the-art results.

Contribution

MOS combines adapter merging and a self-refined retrieval mechanism to effectively mitigate both parameter and retrieval-level forgetting in PTMs for CIL.

Findings

MOS outperforms existing methods on seven benchmarks.

The adapter merging approach preserves task-specific information.

The self-refined retrieval improves adapter selection during inference.

Abstract

Class-Incremental Learning (CIL) requires models to continually acquire knowledge of new classes without forgetting old ones. Despite Pre-trained Models (PTMs) have shown excellent performance in CIL, catastrophic forgetting still occurs as the model learns new concepts. Existing work seeks to utilize lightweight components to adjust the PTM, while the forgetting phenomenon still comes from {\em parameter and retrieval} levels. Specifically, iterative updates of the model result in parameter drift, while mistakenly retrieving irrelevant modules leads to the mismatch during inference. To this end, we propose MOdel Surgery (MOS) to rescue the model from forgetting previous knowledge. By training task-specific adapters, we continually adjust the PTM to downstream tasks. To mitigate parameter-level forgetting, we present an adapter merging approach to learn task-specific adapters, which aims to bridge the gap between different components while reserve task-specific information. Besides, to address retrieval-level forgetting, we introduce a training-free self-refined adapter retrieval mechanism during inference, which leverages the model's inherent ability for better adapter retrieval. By jointly rectifying the model with those steps, MOS can robustly resist catastrophic forgetting in the learning process. Extensive experiments on seven benchmark datasets validate MOS's state-of-the-art performance. Code is available at: https://github.com/sun-hailong/AAAI25-MOS