Transfer Learning Approach for Human Activity Recognition Based on Continuous Wavelet Transform

Author(s)
Olena Pavliuk, Myroslav Mishchuk, Christine Strauss
Abstract

Over the last few years, human activity recognition (HAR) has drawn increasing interest from the scientific community. This attention is mainly attributable to the proliferation of wearable sensors and the expanding role of HAR in such fields as healthcare, sports, and human activity monitoring. Convolutional neural networks (CNN) are becoming a popular approach for addressing HAR problems. However, this method requires extensive training datasets to perform adequately on new data. This paper proposes a novel deep learning model pre-trained on scalograms generated using the continuous wavelet transform (CWT). Nine popular CNN architectures and different CWT configurations were considered to select the best performing combination, resulting in the training and evaluation of more than 300 deep learning models. On the source KU-HAR dataset, the selected model achieved classification accuracy and an F1 score of 97.48% and 97.52%, respectively, which outperformed contemporary state-of-the-art works where this dataset was employed. On the target UCI-HAPT dataset, the proposed model resulted in a maximum accuracy and F1-score increase of 0.21% and 0.33%, respectively, on the whole UCI-HAPT dataset and of 2.82% and 2.89%, respectively, on the UCI-HAPT subset. It was concluded that the usage of the proposed model, particularly with frozen layers, results in improved performance, faster training, and smoother gradient descent on small HAR datasets. However, the use of the pre-trained model on sufficiently large datasets may lead to negative transfer and accuracy degradation.

Organisation(s)
Department of Marketing and International Business
External organisation(s)
Silesian University of Technology, Lviv Polytechnic National University
Journal
Algorithms
Volume
16
ISSN
1999-4893
DOI
https://doi.org/10.3390/a16020077
Publication date
02-2023
Peer reviewed
Yes
Austrian Fields of Science 2012
206001 Biomedical engineering
Keywords
ASJC Scopus subject areas
Computational Mathematics, Theoretical Computer Science, Numerical Analysis, Computational Theory and Mathematics
Sustainable Development Goals
SDG 3 - Good Health and Well-being
Portal url
https://ucrisportal.univie.ac.at/en/publications/transfer-learning-approach-for-human-activity-recognition-based-on-continuous-wavelet-transform(eff98c2f-14eb-409e-b34b-4dbbce7657d5).html