Scene-Adaptive Continual Learning for CSI-based Human Activity Recognition with Mixture of Experts

TL;DR

This paper introduces SAMoE-C, a scalable continual learning method for CSI-based human activity recognition that adapts to different scenes with low inference cost and minimal training overhead.

Contribution

The paper proposes a novel scene-adaptive mixture of experts model with a semantic router and lightweight training protocol for efficient cross-domain CSI-based HAR.

Findings

Approaches state-of-the-art accuracy on a four-scene CSI dataset.

Maintains significantly lower inference cost compared to existing methods.

Requires only a tiny replay buffer for training stability.

Abstract

Channel state information (CSI)-based human activity recognition (HAR) is vulnerable to performance degradation under domain shifts across varying physical environments. Continual learning (CL) offers a principled way to learn new domains sequentially while preserving past knowledge, but existing CL solutions for CSI-based HAR scale poorly with accumulating domains, rely on a large replay buffer, or incur linearly growing inference cost. In this letter, we propose Scene-Adaptive Mixture of Experts with Clustered Specialists (SAMoE-C), which formulates cross-domain CSI-based HAR as a mixture-of-experts system that enables scene-specific adaptation, via an attention-based semantic router that activates only selected experts for each input. Moreover, we develop a novel training protocol, which requires only a tiny replay buffer for stabilizing domain discrimination of the router. Experimental results on a four-scene CSI dataset demonstrate that SAMoE-C approaches the state-of-the-art accuracy, while maintaining a significantly lower inference cost. By jointly combining modular experts, selective activation with router and a lightweight training protocol, SAMoE-C enables scalable cross-domain CSI-based HAR deployment with low training overhead and high computational efficiency in real-world settings.