Inertial Signal Forecasting With Foundation Model Techniques

doi:10.1109/JSEN.2025.3603456

Inertial Signal Forecasting With Foundation Model Techniques

2025-10-01

0

OA

AI

DOI：10.1109/JSEN.2025.3603456

Original

Original request for help

Save

Abstract

En 中文

This article introduces dual-view foundation model (FM), a generative pretrained FM specifically designed for forecasting human inertial signals from body-worn sensors. Unlike traditional transformer-based approaches, dual-view FM employs a novel architecture that mixes global and local signal properties in the time and feature domains to build a comprehensive understanding of inertial activity data. The global view identifies general trends in human inertial signals, while the local view addresses their spontaneous nature by individually analyzing short, sensor-specific signal frames. Our evaluation on eight established datasets shows that dual-view FM accurately forecasts human inertial signals from previously unseen test datasets that involve various sensors, activities, and users. The pretrained models even outperform a production-ready FM for time series forecasting by up to 36 % in mean absolute error (MAE), demonstrating the advantages of domain-specific FMs. Fine-tuning further enhances dual-view FM’s forecasting capabilities so that it surpasses other state-of-the-art architectures by more than 7 % in MAE.

Keywords:

Activity generalization

context awareness

cross-dataset

foundation model (FM)

human activity recognition (HAR)

inertial sensors

self-supervised learning

time series forecasting

Inertial Signal Forecasting With Foundation Model Techniques

Abstract

Journal

Organization