Human activity recognition with deep metric learners.
Martin, Kyle; Wijekoon, Anjana; Wiratunga, Nirmalie
Professor Nirmalie Wiratunga firstname.lastname@example.org
Establishing a strong foundation for similarity-based return is a top priority in Case-Based Reasoning (CBR) systems. Deep Metric Learners (DMLs) are a group of neural network architectures which learn to optimise case representations for similarity-based return by training upon multiple cases simultaneously to incorporate relationship knowledge. This is particularly important in the Human Activity Recognition (HAR) domain, where understanding similarity between cases supports aspects such as personalisation and open-ended HAR. In this paper, we perform a short review of three DMLs and compare their performance across three HAR datasets. Our findings support research which indicates DMLs are valuable to improve similarity-based return and indicate that considering more cases simultaneously offers better performance.
|Start Date||Sep 8, 2019|
|Publication Date||Mar 4, 2020|
|Publisher||CEUR Workshop Proceedings|
|Series Title||CEUR workshop proceedings|
|Institution Citation||MARTIN, K., WIJEKOON, A. and WIRATUNGA, N. 2019. Human activity recognition with deep metric learners. In Kapetanakis, S. and Borck, H. (eds.) Proceedings of 27th International conference on case-based reasoning workshop (ICCBR-WS19), co-located with the 27th International conference on case-based reasoning (ICCBR19), 8-12 September 2019, Otzenhausen, Germany. CEUR Workshop Proceedings, 2567. Aachen: CEUR-WS [online], pages 8-17. Available from: http://ceur-ws.org/Vol-2567/paper1.pdf|
|Keywords||Human activity recognition; Deep metric learning; Deep learning; Metric learning; Matching networks|
MARTIN 2019 Human activity
You might also like
Evaluating the transferability of personalised exercise recognition models.
Learning to recognise exercises for the self-management of low back pain.
A knowledge-light approach to personalised and open-ended human activity recognition.
Developing a catalogue of explainability methods to support expert and non-expert users.