Paper Info
Title
OCEAN: a new opportunistic computing model for wearable activity recognition.
Abstract
Activities of Daily Living (ADL) recognition through wearable devices is an emerging research field. While, for many applications, recognition methods are faced with simultaneously dynamic changes in feature dimension, activity class and data distribution. Existing approaches mainly handle at most one of these three challenges, which significantly affects their performance. In this paper, we propose an Opportunistic Computing model for wEarable Activity recognitioN (OCEAN); by fusing random mapping, fuzzy clustering, and weight updating techniques, OCEAN can online adaptively adjust Single-hidden Layer Feedforward neural network's connection, structure and weight in a coherent manner. Experimental evaluations demonstrate that OCEAN improves the recognition accuracy by 5% to 15% compared to traditional approaches towards dynamic changes.
Year
DOI
Venue
2016
10.1145/2968219.2971453
UbiComp Adjunct
Keywords
Field
DocType
Activity recognition, online learning, opportunistic computing, neural network
Fuzzy clustering,Feedforward neural network,Activity recognition,Random mapping,Computer science,Wearable computer,Artificial intelligence,Artificial neural network,Wearable technology,Machine learning,Feature Dimension
Conference
Citations 
PageRank 
References 
7
0.48
8
Authors
4
Name
Order
Citations
PageRank
Yiqiang Chen11446109.32
Yang Gu2567.54
Xinlong Jiang37610.70
Jindong Wang424716.56