Enhancing Online Continual Learning with Plug-and-Play State Space Model and Class-Conditional Mixture of Discretization

TL;DR

This paper introduces S6MOD, a plug-and-play module for online continual learning that enhances model adaptability by dynamically selecting discretization methods, leading to improved performance and state-of-the-art results.

Contribution

The paper proposes S6MOD, a novel adaptable module that can be integrated into existing models to improve their ability to learn from online data streams.

Findings

S6MOD significantly improves adaptability in online continual learning.

Models with S6MOD achieve state-of-the-art performance.

The module effectively selects the most sensitive discretization method.

Abstract

Online continual learning (OCL) seeks to learn new tasks from data streams that appear only once, while retaining knowledge of previously learned tasks. Most existing methods rely on replay, focusing on enhancing memory retention through regularization or distillation. However, they often overlook the adaptability of the model, limiting the ability to learn generalizable and discriminative features incrementally from online training data. To address this, we introduce a plug-and-play module, S6MOD, which can be integrated into most existing methods and directly improve adaptability. Specifically, S6MOD introduces an extra branch after the backbone, where a mixture of discretization selectively adjusts parameters in a selective state space model, enriching selective scan patterns such that the model can adaptively select the most sensitive discretization method for current dynamics. We further design a class-conditional routing algorithm for dynamic, uncertainty-based adjustment and implement a contrastive discretization loss to optimize it. Extensive experiments combining our module with various models demonstrate that S6MOD significantly enhances model adaptability, leading to substantial performance gains and achieving the state-of-the-art results.