Bridging Generalization and Personalization in Human Activity Recognition via On-Device Few-Shot Learning

TL;DR

This paper introduces an on-device few-shot learning framework for human activity recognition that effectively combines generalization across users with rapid personalization, enabling accurate and energy-efficient wearable activity recognition on resource-limited devices.

Contribution

The paper presents a novel on-device few-shot learning method that trains a generalizable model and quickly adapts to new users with minimal data, suitable for deployment on microcontrollers.

Findings

Improved accuracy by up to 17.38% after adaptation.

Achieved robust on-device learning with minimal computation.

Demonstrated effectiveness on three benchmark datasets.

Abstract

Human Activity Recognition (HAR) with different sensing modalities requires both strong generalization across diverse users and efficient personalization for individuals. However, conventional HAR models often fail to generalize when faced with user-specific variations, leading to degraded performance. To address this challenge, we propose a novel on-device few-shot learning framework that bridges generalization and personalization in HAR. Our method first trains a generalizable representation across users and then rapidly adapts to new users with only a few labeled samples, updating lightweight classifier layers directly on resource-constrained devices. This approach achieves robust on-device learning with minimal computation and memory cost, making it practical for real-world deployment. We implement our framework on the energy-efficient RISC-V GAP9 microcontroller and evaluate it on three benchmark datasets (RecGym, QVAR-Gesture, Ultrasound-Gesture). Across these scenarios, post-deployment adaptation improves accuracy by 3.73\%, 17.38\%, and 3.70\%, respectively. These results demonstrate that few-shot on-device learning enables scalable, user-aware, and energy-efficient wearable human activity recognition by seamlessly uniting generalization and personalization. The related framework is open sourced for further research\footnote{https://github.com/kangpx/onlineTiny2023}.