Affiliations: School of EECS, University of Central Florida, Orlando, FL, USA | Carnegie Mellon University, Institute for Complex Engineered Systems, Department of Electrical and Computer Engineering, Pittsburgh, PA 15213, USA
Correspondence:
[*]
Address for correspondence: Dr. Gita Sukthankar, School of Electrical Engineering and Computer Science, University of Central Florida, 4000 Central Florida Blvd, Orlando, FL 32816-2362, USA. Tel.: +1 407 823 4305; Fax: +1 407 823 2762; E-mail: gitars@eecs.ucf.edu.
Abstract: The ability to accurately recognize human household activities is an important stepping stone toward creating home living assistance systems in the future. Classifying these activities can be difficult due to noisy sensor data, lack of labeled training samples for rare actions and large individual differences in activity execution. In this article, we present two techniques for improving the supervised classification of human activities from motion data: 1) an active learning framework to improve sample efficiency and 2) intelligent feature selection to reduce training time. We demonstrate our techniques using the CMU Multimodal Activity database.
Keywords: Activity recognition, active learning, feature selection, support vector machines, conditional random fields
