Multi-label classification using label combination to recognize human activity based on various sensor positions

Activity recognition using accelerometer sensor shows the good and positive impact in current health style perseverance. The sensor already built in the various smartphones, belts, and tapes for easy usage and applications in order to detect person’s activity in real time response. Ram-pant research...

Full description

Bibliographic Details
Main Authors: Zainudin, Muhammad Noorazlan Shah, Mohamed, Raihani, Sulaiman, Md. Nasir, Perumal, Thinagaran, Mustapha, Norwati, Ahmad Nazri, Azree Shahrel
Format: Conference or Workshop Item
Language:English
Published: School of Computing, UUM College of Arts and Sciences 2017
Online Access:http://psasir.upm.edu.my/id/eprint/64452/
http://psasir.upm.edu.my/id/eprint/64452/1/PID63-669-674e.pdf
_version_ 1848855006057332736
author Zainudin, Muhammad Noorazlan Shah
Mohamed, Raihani
Sulaiman, Md. Nasir
Perumal, Thinagaran
Mustapha, Norwati
Ahmad Nazri, Azree Shahrel
author_facet Zainudin, Muhammad Noorazlan Shah
Mohamed, Raihani
Sulaiman, Md. Nasir
Perumal, Thinagaran
Mustapha, Norwati
Ahmad Nazri, Azree Shahrel
author_sort Zainudin, Muhammad Noorazlan Shah
building UPM Institutional Repository
collection Online Access
description Activity recognition using accelerometer sensor shows the good and positive impact in current health style perseverance. The sensor already built in the various smartphones, belts, and tapes for easy usage and applications in order to detect person’s activity in real time response. Ram-pant research has been done to measure the effectiveness of accelerometer location detection of the activity. Hence, a separate method has been used to tackle the issues. This paper proposed multi-label classification (MLC) to look the effectiveness of the sensor location with a correlation of types of activity at the same time the similar activity could be discriminated accurately. The traditional single label works using Random Forest (RF) was used and the accuracy of the model will be compared with MLC using LC (Label Combination) with RF as a base classifier. As a result, MLC outperform the traditional single classifier approach to distinguish the accuracy for both stairs activities such as downstairs and upstairs is highly accurate using proposed method. At the same time, the pocket position successfully considered as the best sensor position to recognize various types of activities.
first_indexed 2025-11-15T11:18:53Z
format Conference or Workshop Item
id upm-64452
institution Universiti Putra Malaysia
institution_category Local University
language English
last_indexed 2025-11-15T11:18:53Z
publishDate 2017
publisher School of Computing, UUM College of Arts and Sciences
recordtype eprints
repository_type Digital Repository
spelling upm-644522018-07-05T09:35:20Z http://psasir.upm.edu.my/id/eprint/64452/ Multi-label classification using label combination to recognize human activity based on various sensor positions Zainudin, Muhammad Noorazlan Shah Mohamed, Raihani Sulaiman, Md. Nasir Perumal, Thinagaran Mustapha, Norwati Ahmad Nazri, Azree Shahrel Activity recognition using accelerometer sensor shows the good and positive impact in current health style perseverance. The sensor already built in the various smartphones, belts, and tapes for easy usage and applications in order to detect person’s activity in real time response. Ram-pant research has been done to measure the effectiveness of accelerometer location detection of the activity. Hence, a separate method has been used to tackle the issues. This paper proposed multi-label classification (MLC) to look the effectiveness of the sensor location with a correlation of types of activity at the same time the similar activity could be discriminated accurately. The traditional single label works using Random Forest (RF) was used and the accuracy of the model will be compared with MLC using LC (Label Combination) with RF as a base classifier. As a result, MLC outperform the traditional single classifier approach to distinguish the accuracy for both stairs activities such as downstairs and upstairs is highly accurate using proposed method. At the same time, the pocket position successfully considered as the best sensor position to recognize various types of activities. School of Computing, UUM College of Arts and Sciences 2017 Conference or Workshop Item PeerReviewed text en http://psasir.upm.edu.my/id/eprint/64452/1/PID63-669-674e.pdf Zainudin, Muhammad Noorazlan Shah and Mohamed, Raihani and Sulaiman, Md. Nasir and Perumal, Thinagaran and Mustapha, Norwati and Ahmad Nazri, Azree Shahrel (2017) Multi-label classification using label combination to recognize human activity based on various sensor positions. In: 6th International Conference on Computing and Informatics (ICOCI 2017), 25-27 Apr. 2017, Kuala Lumpur, Malaysia. (pp. 669-674).
spellingShingle Zainudin, Muhammad Noorazlan Shah
Mohamed, Raihani
Sulaiman, Md. Nasir
Perumal, Thinagaran
Mustapha, Norwati
Ahmad Nazri, Azree Shahrel
Multi-label classification using label combination to recognize human activity based on various sensor positions
title Multi-label classification using label combination to recognize human activity based on various sensor positions
title_full Multi-label classification using label combination to recognize human activity based on various sensor positions
title_fullStr Multi-label classification using label combination to recognize human activity based on various sensor positions
title_full_unstemmed Multi-label classification using label combination to recognize human activity based on various sensor positions
title_short Multi-label classification using label combination to recognize human activity based on various sensor positions
title_sort multi-label classification using label combination to recognize human activity based on various sensor positions
url http://psasir.upm.edu.my/id/eprint/64452/
http://psasir.upm.edu.my/id/eprint/64452/1/PID63-669-674e.pdf