Multi-resident activity recognition using incremental decision trees

Authors: Prossegger, M. and Bouchachia, A.

Journal: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Volume: 8779 LNAI

Pages: 182-191

Publisher: Springer Verlag

eISSN: 1611-3349

ISBN: 9783319112978

ISSN: 0302-9743

DOI: 10.1007/978-3-319-11298-5-19

Abstract:

The present paper proposes the application of decision trees to model activities of daily living in a multi-resident context. An extension of ID5R, called E-ID5R, is proposed. It augments the leaf nodes and allows such nodes to be multi-labeled. E-ID5R induces a decision tree incrementally to accommodate new instances and new activities as they become available over time. To evaluate the proposed algorithm, the ARAS dataset which is a real-world multi-resident dataset stemming from two houses is used. E-ID5R performs differently on activities of both houses. © 2014 Springer International Publishing Switzerland.

Source: Scopus

Preferred by: Hamid Bouchachia