Make Continual Learning Stronger via C-Flat

TL;DR

This paper introduces C-Flat, a simple and effective method to improve continual learning by promoting flatter loss landscapes, which enhances model generalization across sequential tasks.

Contribution

The paper proposes a novel, easy-to-implement flatness-based regularization method for continual learning, compatible with various CL approaches and improving performance.

Findings

C-Flat improves CL performance in most cases.

It is a plug-and-play method requiring only one line of code.

C-Flat enhances model generalization by promoting flatter minima.

Abstract

Model generalization ability upon incrementally acquiring dynamically updating knowledge from sequentially arriving tasks is crucial to tackle the sensitivity-stability dilemma in Continual Learning (CL). Weight loss landscape sharpness minimization seeking for flat minima lying in neighborhoods with uniform low loss or smooth gradient is proven to be a strong training regime improving model generalization compared with loss minimization based optimizer like SGD. Yet only a few works have discussed this training regime for CL, proving that dedicated designed zeroth-order sharpness optimizer can improve CL performance. In this work, we propose a Continual Flatness (C-Flat) method featuring a flatter loss landscape tailored for CL. C-Flat could be easily called with only one line of code and is plug-and-play to any CL methods. A general framework of C-Flat applied to all CL categories and a thorough comparison with loss minima optimizer and flat minima based CL approaches is presented in this paper, showing that our method can boost CL performance in almost all cases. Code is available at https://github.com/WanNaa/C-Flat.