Activity Recognition Based on SVM Kernel Fusion in Smart Home
Smart home is regarded as an independent healthy living for elderly person and it demands active research in activity recognition. This paper proposes kernel fusion method, using Support Vector Machine (SVM) in order to improve the accuracy of performed a
- PDF / 374,947 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 28 Downloads / 180 Views
Abstract Smart home is regarded as an independent healthy living for elderly person and it demands active research in activity recognition. This paper proposes kernel fusion method, using Support Vector Machine (SVM) in order to improve the accuracy of performed activities. Although, SVM is a powerful statistical technique, but still suffer from the expected level of accuracy due to complex feature space. Designing a new kernel function is difficult task, while common available kernel functions are not adequate for the activity recognition domain to achieve high accuracy. We introduce a method, to train the different SVMs independently over the standard kernel functions and fuse the individual results on the decision level to increase the confidence of each activity class. The proposed approach has been evaluated on ten different kinds of activities from CASAS smart home (Tulum 2009) real world dataset. We compare our SVM kernel fusion approach with the standard kernel functions and get overall accuracy of 91.41 %. Keywords Activity recognition
Smart home SVM Kernel fusion
M. Fahim (&) I. Fatima S. Lee Y.-K. Lee Ubiquitos Computing Laboratory, Department of Comptuer Engineering, Kyung Hee University, Suwon, Korea e-mail: [email protected] I. Fatima e-mail: [email protected] S. Lee e-mail: [email protected] Y.-K. Lee e-mail: [email protected]
S.-S. Yeo et al. (eds.), Computer Science and its Applications, Lecture Notes in Electrical Engineering 203, DOI: 10.1007/978-94-007-5699-1_29, Ó Springer Science+Business Media Dordrecht 2012
283
284
M. Fahim et al.
1 Introduction Activity recognition is active area of research since the inclusion of smart home concept for providing ubiquitous lifecare services. It can provide a valuable health monitoring services for ageing society to improve their quality of life. In the recent years, several smart homes have been developed such as CASAS and MavHome [1] at Washington State University, Aware Home [2] at Georgia Tech University, Adaptive House [3] at University of Colorado, House_n [4] at Massachusetts Institute of Technology (MIT), and House A [5] at Intelligent Systems Laboratory. The nomenclature of the activity recognition has two broad categories, visual sensing and ubiquitous sensing technologies. In first category, camera based techniques are used to monitor the behavior of inhabitants. These are not practical due to privacy reason, day/night vision problem and jumble environment. In second category, ubiquitous sensors are embedded on the different objects to recognize the daily life activities. They are cost-effective, easy to install and less intrusive to the privacy of inhabitants. Activity recognition is a big challenge by using ubiquitous sensors due to complex and highly diverse life styles. Several machine learning and statistical approaches have been proposed to recognize the human activities and achieves acceptable accuracy with particular attentions [6, 7]. This paper investigates the theoretically rich statistical met
Data Loading...
