Elara
  • Home
  • Research
  • Projects
  • CV
  • Home
  • Research
  • Projects
  • CV
GitHubLinkedInInstagramORCIDBlueskyRSS Feed

© 2025 Elara Liu | All rights reserved.

Themed by EnjuFolio · Crafted by Elara Liu

Smartwatch IMU foundation model for human activity recognition

Aidan Bradshaw, Zhuoran Liu, Riku Arakawa, Arnav Choudhry, Xin Liu, Karan Ahuja

Advised by:Dr. Karan Ahuja (Northwestern University)

ICML 2026May, 2025

Keywords:
Smartwatch IMUHuman activity recognitionTime-series foundation modelsSelf-supervised learningDomain generalizationWearable sensing

Abstract

I am co-developing a smartwatch IMU foundation model that aggregates and harmonizes multiple datasets, then pretrains self-supervised sequence models and tests them under leave-one-dataset and unseen-device protocols, aiming for a generalizable, well-documented model and recipe that transfers across devices, populations, and everyday conditions.

1 Overview

In Prof. Ahuja’s lab, I am tackling a persistent gap in human activity recognition: models trained on a single IMU dataset often look impressive on their home benchmark yet fall apart when the watch, sampling rate, or population changes. To push against that brittleness, I started by aggregating and harmonizing multiple public smartwatch datasets, aligning device frames, reconciling sampling rates, and negotiating a shared label ontology that can express each dataset’s idiosyncratic activity set without collapsing everything into vague catch-alls. A lot of the early work was spent iterating on these choices—discovering where alignment conventions disagreed, where resampling quietly distorted motion patterns, and where label merges hid important distinctions—until simple baselines could at least behave sensibly across held-out datasets.

On top of this common substrate, I am pretraining large sequence models—transformer-style architectures and time-series variants of vision models—using self-supervised objectives such as masked modeling and contrastive invariance. The evaluation pipeline is explicitly designed to stress generalization: leave-one-dataset-out and unseen-device protocols, plus planned ablations that vary window length, sampling rate, augmentation, and sensor stacks to isolate which design decisions actually move the needle. When early results exposed failure modes on particular datasets, I treated them as prompts to revisit both the harmonization and the objectives rather than just tuning hyperparameters, aiming for a training recipe that is robust, reproducible, and well-documented. The long-term goal is a portable IMU foundation model that researchers and practitioners can fine-tune across devices and populations, with clear guidelines for adapting it to wellness and community settings instead of yet another leaderboard-only model.