Aggregate Features in Multisample Classification Problems

Author

Varga, Robert ; Matheson, S. Marie ; Hamilton-Wright, Andrew

Author_Institution

Sch. of Comput. Sci., Univ. of Guelph, Guelph, ON, Canada

Volume

19

Issue

2

fYear

2015

fDate

Mar-15

Firstpage

486

Lastpage

492

Abstract

This paper evaluates the classification of multisample problems, such as electromyographic (EMG) data, by making aggregate features available to a per-sample classifier. It is found that the accuracy of this approach is superior to that of traditional methods such as majority vote for this problem. The classification improvements of this method, in conjunction with a confidence measure expressing the per-sample probability of classification failure (i.e., a hazard function) is described and measured. Results are expected to be of interest in clinical decision support system development.

Keywords

Bayes methods; data analysis; electromyography; pattern classification; EMG data; aggregate feature; classification failure per-sample probability; classification improvement; electromyographic data; multisample problem classification; per-sample classifier; Accuracy; Bayes methods; Biomedical measurement; Informatics; Labeling; Muscles; Training data; Bayes methods; decision support systems; machine learning; pattern analysis; statistical learning; supervised learning;

fLanguage

English

Journal_Title

Biomedical and Health Informatics, IEEE Journal of

Publisher

ieee

ISSN

2168-2194

Type

jour

DOI

10.1109/JBHI.2014.2314856

Filename

6782399