Exploring Contrastive Learning in Human Activity Recognition for Healthcare

TL;DR

This paper investigates the application of contrastive learning, specifically SimCLR, to human activity recognition in healthcare, showing potential improvements over traditional methods through extensive data augmentation analysis.

Contribution

It adapts SimCLR contrastive learning to HAR, providing an extensive evaluation of data augmentation techniques and their impact on model performance in healthcare contexts.

Findings

Contrastive learning can improve HAR performance with proper data augmentation.

Data augmentation order and type significantly affect results.

Fine-tuning with rotation augmentation shows promising improvements.

Abstract

Human Activity Recognition (HAR) constitutes one of the most important tasks for wearable and mobile sensing given its implications in human well-being and health monitoring. Motivated by the limitations of labeled datasets in HAR, particularly when employed in healthcare-related applications, this work explores the adoption and adaptation of SimCLR, a contrastive learning technique for visual representations, to HAR. The use of contrastive learning objectives causes the representations of corresponding views to be more similar, and those of non-corresponding views to be more different. After an extensive evaluation exploring 64 combinations of different signal transformations for augmenting the data, we observed significant performance differences owing to the order and the function thereof. In particular, preliminary results indicated an improvement over supervised and unsupervised learning methods when using fine-tuning and random rotation for augmentation, however, future work should explore under which conditions SimCLR is beneficial for HAR systems and other healthcare-related applications.