Evaluating the transferability of personalised exercise recognition models.

Wijekoon, Anjana; Wiratunga, Nirmalie

Abstract

Exercise Recognition (ExR) is relevant in many high impact domains, from health care to recreational activities to sports sciences. Like Human Activity Recognition (HAR), ExR faces many challenges when deployed in the real-world. For instance, typical lab performances of Machine Learning models, are hard to replicate, due to differences in personal nuances, traits and ambulatory rhythms. Thus effective transferability of a trained ExR model, depends on its ability to adapt and personalise to new users or user groups. This calls for new experimental design strategies that are also person-aware, and able to organise train and test data differently from standard ML practice. Speciffically, we look at person-agnostic and person-aware methods of train-test data creation, and compare them to identify best practices on a comparative study of personalised ExR model transfer. Our findings show that ExR when compared to results with other HAR tasks, to be a far more challenging personalisation problem and also confirms the utility of metric learning algorithms for personalised model transfer.

Citation

WIJEKOON, A. and WIRATUNGA, N. 2020. Evaluating the transferability of personalised exercise recognition models. In Iliadis, L., Angelov, P.P., Jayne, C. and Pimenidis, E. (eds.) Proceedings of the 21st Engineering applications of neural networks conference 2020 (EANN 2020): proceedings of the EANN 2020, 5-7 June 2020, Halkidiki, Greece. Proceedings of the International Neural Networks Society, 2. Cham: Springer [online], pages 32-44. Available from: https://doi.org/10.1007/978-3-030-48791-1_3

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 21st International engineering applications of neural networks conference (EANN 2020)
• Start Date: Jun 5, 2020
• End Date: Jun 7, 2020
• Acceptance Date: Mar 29, 2020
• Online Publication Date: May 28, 2020
• Publication Date: Dec 31, 2020
• Deposit Date: Apr 2, 2020
• Publicly Available Date: May 29, 2021
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Volume: 16268
• Pages: 32-44
• Series Title: Proceedings of the International Neural Networks Society
• Series Number: 2
• Series ISSN: 2661-8141
• Book Title: Proceedings of the 21st Engineering applications of neural networks conference 2020 (EANN 2020): proceedings of the EANN 2020
• ISBN: 9783030487904
• DOI: https://doi.org/10.1007/978-3-030-48791-1_3
• Keywords: Exercise recognition; Transferability; Personalisation; Performance evaluation
• Public URL: https://rgu-repository.worktribe.com/output/888634