Distribution-based semi-supervised learning for activity recognition

Supervised learning methods have been widely applied to activity recognition. The prevalent success of existing methods, however, has two crucial prerequisites: proper feature extraction and sufficient labeled training data. The former is important to differentiate activities, while the latter is cr...

Full description

Saved in:
Bibliographic Details
Main Authors: Qian, Hangwei, Pan, Sinno Jialin, Miao, Chunyan
Other Authors: School of Computer Science and Engineering
Format: Conference or Workshop Item
Language:English
Published: 2019
Subjects:
Online Access:https://hdl.handle.net/10356/106004
http://hdl.handle.net/10220/49629
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-106004
record_format dspace
spelling sg-ntu-dr.10356-1060042022-07-21T08:19:29Z Distribution-based semi-supervised learning for activity recognition Qian, Hangwei Pan, Sinno Jialin Miao, Chunyan School of Computer Science and Engineering Interdisciplinary Graduate School (IGS) The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19) Joint NTU-UBC Research Centre of Excellence in Active Living for the Elderly (LILY) Wireless-sensor-based Engineering::Computer science and engineering Semi-supervised Learning Supervised learning methods have been widely applied to activity recognition. The prevalent success of existing methods, however, has two crucial prerequisites: proper feature extraction and sufficient labeled training data. The former is important to differentiate activities, while the latter is crucial to build a precise learning model. These two prerequisites have become bottlenecks to make existing methods more practical. Most existing feature extraction methods highly depend on domain knowledge, while labeled data requires intensive human annotation effort. Therefore, in this paper, we propose a novel method, named Distribution-based Semi-Supervised Learning, to tackle the aforementioned limitations. The proposed method is capable of automatically extracting powerful features with no domain knowledge required, meanwhile, alleviating the heavy annotation effort through semi-supervised learning. Specifically, we treat data stream of sensor readings received in a period as a distribution, and map all training distributions, including labeled and unlabeled, into a reproducing kernel Hilbert space (RKHS) using the kernel mean embedding technique. The RKHS is further altered by exploiting the underlying geometry structure of the unlabeled distributions. Finally, in the altered RKHS, a classifier is trained with the labeled distributions. We conduct extensive experiments on three public datasets to verify the effectiveness of our method compared with state-of-the-art baselines. 2019-08-14T06:46:29Z 2019-12-06T22:02:42Z 2019-08-14T06:46:29Z 2019-12-06T22:02:42Z 2019 Conference Paper Qian, H., Pan, S. J., & Miao, C. (2019). Distribution-based semi-supervised learning for activity recognition. The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19). https://hdl.handle.net/10356/106004 http://hdl.handle.net/10220/49629 en © 2019 Association for the Advancement of Artificial Intelligence (AAAI). All rights reserved. This paper was published in The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19) and is made available with permission of Association for the Advancement of Artificial Intelligence (AAAI). 8 p. application/pdf
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Wireless-sensor-based
Engineering::Computer science and engineering
Semi-supervised Learning
spellingShingle Wireless-sensor-based
Engineering::Computer science and engineering
Semi-supervised Learning
Qian, Hangwei
Pan, Sinno Jialin
Miao, Chunyan
Distribution-based semi-supervised learning for activity recognition
description Supervised learning methods have been widely applied to activity recognition. The prevalent success of existing methods, however, has two crucial prerequisites: proper feature extraction and sufficient labeled training data. The former is important to differentiate activities, while the latter is crucial to build a precise learning model. These two prerequisites have become bottlenecks to make existing methods more practical. Most existing feature extraction methods highly depend on domain knowledge, while labeled data requires intensive human annotation effort. Therefore, in this paper, we propose a novel method, named Distribution-based Semi-Supervised Learning, to tackle the aforementioned limitations. The proposed method is capable of automatically extracting powerful features with no domain knowledge required, meanwhile, alleviating the heavy annotation effort through semi-supervised learning. Specifically, we treat data stream of sensor readings received in a period as a distribution, and map all training distributions, including labeled and unlabeled, into a reproducing kernel Hilbert space (RKHS) using the kernel mean embedding technique. The RKHS is further altered by exploiting the underlying geometry structure of the unlabeled distributions. Finally, in the altered RKHS, a classifier is trained with the labeled distributions. We conduct extensive experiments on three public datasets to verify the effectiveness of our method compared with state-of-the-art baselines.
author2 School of Computer Science and Engineering
author_facet School of Computer Science and Engineering
Qian, Hangwei
Pan, Sinno Jialin
Miao, Chunyan
format Conference or Workshop Item
author Qian, Hangwei
Pan, Sinno Jialin
Miao, Chunyan
author_sort Qian, Hangwei
title Distribution-based semi-supervised learning for activity recognition
title_short Distribution-based semi-supervised learning for activity recognition
title_full Distribution-based semi-supervised learning for activity recognition
title_fullStr Distribution-based semi-supervised learning for activity recognition
title_full_unstemmed Distribution-based semi-supervised learning for activity recognition
title_sort distribution-based semi-supervised learning for activity recognition
publishDate 2019
url https://hdl.handle.net/10356/106004
http://hdl.handle.net/10220/49629
_version_ 1739837430737928192