Machine Learning Techniques for Sensor-based Human Activity Recognition with Data Heterogeneity -- A Review

TL;DR

This review examines how machine learning techniques tackle data heterogeneity in sensor-based human activity recognition, highlighting challenges, methods, datasets, and future directions to improve model adaptability and performance.

Contribution

It categorizes data heterogeneity types in HAR and reviews machine learning approaches tailored to each, providing a comprehensive overview of current methods and future challenges.

Findings

Different data heterogeneity types require specific machine learning strategies.

Addressing data heterogeneity improves HAR model performance and personalization.

The review summarizes key datasets and discusses future research directions.

Abstract

Sensor-based Human Activity Recognition (HAR) is crucial in ubiquitous computing, analysing behaviours through multi-dimensional observations. Despite research progress, HAR confronts challenges, particularly in data distribution assumptions. Most studies often assume uniform data distributions across datasets, contrasting with the varied nature of practical sensor data in human activities. Addressing data heterogeneity issues can improve performance, reduce computational costs, and aid in developing personalized, adaptive models with less annotated data. This review investigates how machine learning addresses data heterogeneity in HAR, by categorizing data heterogeneity types, applying corresponding suitable machine learning methods, summarizing available datasets, and discussing future challenges.