@inproceedings{8858aa83ea994b9c947119d2b448a2e7,
title = "Physical activity monitoring with mobile phones",
abstract = "The rich sensing ability of smart mobile phones brings an unique opportunity to detect and long-term monitor people's physical activities. However, with mobile phone the application has to comply with people's usage habit of it and thus capture the right moment to recognize activities, which will potentially cause great in-class variances. As a result, the model potentially becomes complex and costs much computing resources in mobile phone. This paper recognize people's physical activities when they place the mobile phone in the pockets near the pelvic region. Experiment results show that the accuracy could reach 97.7\%. To reduce the model size, evaluation of each feature attribution contribution for the accuracy is performed. And the result shows that we can cut the feature dimension from 22 to 8 while obtaining the smallest model.",
keywords = "accelerometer, activity recognition, feature reduction, machine learning, mobile phone",
author = "Lin Sun and Daqing Zhang and Nan Li",
year = "2011",
month = jun,
day = "30",
doi = "10.1007/978-3-642-21535-3\_14",
language = "English",
isbn = "9783642215346",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "104--111",
booktitle = "Toward Useful Services for Elderly and People with Disabilities - 9th International Conference on Smart Homes and Health Telematics, ICOST 2011, Proceedings",
note = "9th International Conference on Smart Homes and Health Telematics: Toward Useful Services for Elderly and People with Disabilities, ICOST 2011 ; Conference date: 20-06-2011 Through 22-06-2011",
}