Classification of physical activities based on sparse representation

Author

Shaopeng Liu ; Gao, Robert X. ; John, Deepak ; Staudenmayer, J. ; Freedson, P.S.

Author_Institution

Dept. of Mech. Eng., Univ. of Connecticut, Storrs, CT, USA

fYear

2012

fDate

Aug. 28 2012-Sept. 1 2012

Firstpage

6200

Lastpage

6203

Abstract

This paper presents a new classification method for physical activity assessment, based on sparse representation. This method bypasses the need for feature extraction and selection that is typically involved for activity classification, and classifies activities using raw sensor signals directly. Higher discriminative power than that from the conventional k-nearest neighbor algorithm has been demonstrated through experiments performed on 105 subjects.

Keywords

biomechanics; feature extraction; medical signal processing; signal classification; signal representation; Higher discriminative power; classification method; conventional k-nearest neighbor algorithm; feature extraction; feature selection; physical activity assessment; raw sensor signals; sparse representation; Accuracy; Classification algorithms; Feature extraction; Legged locomotion; Sparse matrices; Training; Vectors; Algorithms; Humans; Motor Activity;

fLanguage

English

Publisher

ieee

Conference_Titel

Engineering in Medicine and Biology Society (EMBC), 2012 Annual International Conference of the IEEE

Conference_Location

San Diego, CA

ISSN

1557-170X

Print_ISBN

978-1-4244-4119-8

Electronic_ISBN

1557-170X

Type

conf

DOI

10.1109/EMBC.2012.6347410

Filename

6347410