Efficient Deep Clustering of Human Activities and How to Improve Evaluation

Louis Mahon, Thomas Lukasiewicz
Proceedings of The 14th Asian Conference on Machine Learning, PMLR 189:722-737, 2023.

Abstract

There has been much recent research on human activity recognition (HAR), due to the proliferation of wearable sensors in watches and phones, and the advances of deep learning methods, which avoid the need to manually extract features from raw sensor signals. A significant disadvantage of deep learning applied to HAR is the need for manually labelled training data, which is especially difficult to obtain for HAR datasets. Progress is starting to be made in the unsupervised setting, in the form of deep HAR clustering models, which can assign labels to data without having been given any labels to train on, but there are problems with evaluating deep HAR clustering models, which makes assessing the field and devising new methods difficult. In this paper, we highlight several distinct problems with how deep HAR clustering models are evaluated, describing these problems in detail and conducting careful experiments to explicate the effect that they can have on results. Additionally, we present a new deep clustering model for HAR. When tested under our proposed settings, our model performs better than (or on par with) existing models, while also being more efficient and scalable by avoiding the need for an autoencoder.

Cite this Paper


BibTeX
@InProceedings{pmlr-v189-mahon23a, title = {Efficient Deep Clustering of Human Activities and How to Improve Evaluation}, author = {Mahon, Louis and Lukasiewicz, Thomas}, booktitle = {Proceedings of The 14th Asian Conference on Machine Learning}, pages = {722--737}, year = {2023}, editor = {Khan, Emtiyaz and Gonen, Mehmet}, volume = {189}, series = {Proceedings of Machine Learning Research}, month = {12--14 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v189/mahon23a/mahon23a.pdf}, url = {https://proceedings.mlr.press/v189/mahon23a.html}, abstract = {There has been much recent research on human activity recognition (HAR), due to the proliferation of wearable sensors in watches and phones, and the advances of deep learning methods, which avoid the need to manually extract features from raw sensor signals. A significant disadvantage of deep learning applied to HAR is the need for manually labelled training data, which is especially difficult to obtain for HAR datasets. Progress is starting to be made in the unsupervised setting, in the form of deep HAR clustering models, which can assign labels to data without having been given any labels to train on, but there are problems with evaluating deep HAR clustering models, which makes assessing the field and devising new methods difficult. In this paper, we highlight several distinct problems with how deep HAR clustering models are evaluated, describing these problems in detail and conducting careful experiments to explicate the effect that they can have on results. Additionally, we present a new deep clustering model for HAR. When tested under our proposed settings, our model performs better than (or on par with) existing models, while also being more efficient and scalable by avoiding the need for an autoencoder.} }
Endnote
%0 Conference Paper %T Efficient Deep Clustering of Human Activities and How to Improve Evaluation %A Louis Mahon %A Thomas Lukasiewicz %B Proceedings of The 14th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Emtiyaz Khan %E Mehmet Gonen %F pmlr-v189-mahon23a %I PMLR %P 722--737 %U https://proceedings.mlr.press/v189/mahon23a.html %V 189 %X There has been much recent research on human activity recognition (HAR), due to the proliferation of wearable sensors in watches and phones, and the advances of deep learning methods, which avoid the need to manually extract features from raw sensor signals. A significant disadvantage of deep learning applied to HAR is the need for manually labelled training data, which is especially difficult to obtain for HAR datasets. Progress is starting to be made in the unsupervised setting, in the form of deep HAR clustering models, which can assign labels to data without having been given any labels to train on, but there are problems with evaluating deep HAR clustering models, which makes assessing the field and devising new methods difficult. In this paper, we highlight several distinct problems with how deep HAR clustering models are evaluated, describing these problems in detail and conducting careful experiments to explicate the effect that they can have on results. Additionally, we present a new deep clustering model for HAR. When tested under our proposed settings, our model performs better than (or on par with) existing models, while also being more efficient and scalable by avoiding the need for an autoencoder.
APA
Mahon, L. & Lukasiewicz, T.. (2023). Efficient Deep Clustering of Human Activities and How to Improve Evaluation. Proceedings of The 14th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 189:722-737 Available from https://proceedings.mlr.press/v189/mahon23a.html.

Related Material