Reducing the Response Time for Activity Recognition Through use of Prototype Generation Algorithms

详细信息    查看全文

• 关键词：Activity recognition ; Data ; driven ; Nearest Neighbor (NN) ; Prototype Generation (PG) ; Response time
• 刊名：Lecture Notes in Computer Science
• 出版年：2015
• 出版时间：2015
• 年：2015
• 卷：9102
• 期：1
• 页码：313-318
• 全文大小：113 KB
• 参考文献：1.Satyanarayanan, M.: Pervasive computing: Vision and challenges. IEEE Personal Communications 8(4), 10鈥?7 (2001)View Article
  2.Emmanouilidis, C., Koutsiamanis, R.-A., Tasidou, A.: Mobile guides: Taxonomy of architectures, context awareness, technologies and applications. Journal of Network and Computer Applications 36(1), 103鈥?25 (2013)View Article
  3.Alam, M., Hamida, E.: Surveying wearable human assistive technology for life and safety critical applications: Standards, challenges and opportunities. Sensors (Switzerland) 14(5), 9153鈥?209 (2014)View Article
  4.Cover, T., Hart, P.: Nearest neighbor pattern classification. IEEE Transactions on Information Theory 13, 21鈥?7 (1967)View Article MATH
  5.Wu, X., Kumar, V.: The Top Ten Algorithms in Data Mining. Chapman & Hall/CRC, 1st ed. (2009)
  6.Moayeri Pour, G., Troped, P., Evans, J.: Environment feature extraction and classification for context aware physical activity monitoring, pp. 123鈥?28 (2013)
  7.Kononenko, I., Kukar, M.: Machine Learning and Data Mining: Introduction to Principles and Algorithms. Horwood Publishing Limited (2007)
  8.Garcia, S., Derrac, J., Cano, J., Herrera, F.: Prototype selection for nearest neighbor classification: Taxonomy and empirical study. IEEE Transactions on Pattern Analysis and Machine Intelligence 34(3), 417鈥?35 (2012)View Article
  9.Lozano, M., Sotoca, J., Sanchez, J., Pla, F., Pekalska, E., Duin, R.: Experimental study on prototype optimisation algorithms for prototype-based classification in vector spaces. Pattern Recognition 39(10), 1827鈥?838 (2006)View Article MATH
  10.Wilson, R.D., Martinez, T.: Reduction techniques for instance-based learning algorithms. Machine Learning 38(3), 257鈥?86 (2000)View Article MATH
  11.Van Kasteren, T., Noulas, A., Englebienne, G., Krse, B.: Accurate activity recognition in a home setting, pp. 1鈥? (2008)
  12.Alcala Fdez, J., Fernandez, A., Luengo, J., Derrac, J., Garcia, S., Sanchez, L., Herrera, F.: Keel data-mining software tool: Data set repository, integration of algorithms and experimental analysis framework. Journal of Multiple-Valued Logic and Soft Computing 17(2鈥?), 255鈥?87 (2011)
  
• 作者单位：Macarena Espinilla (18)
  Francisco J. Quesada (18)
  Francisco Moya (18)
  Luis Mart铆nez (18)
  Chris D. Nugent (19)
  
  18. Computer Sciences Department, University of Ja茅n, Campus Las Lagunillas s/n, 23071, Ja茅n, Spain
  19. School of Computing and Mathematics, University of Ulster, Jordanstown, Coleraine, BT37 0QB, UK
  
• 丛书名：Inclusive Smart Cities and e-Health
• ISBN：978-3-319-19312-0
• 刊物类别：Computer Science
• 刊物主题：Artificial Intelligence and Robotics
  Computer Communication Networks
  Software Engineering
  Data Encryption
  Database Management
  Computation by Abstract Devices
  Algorithm Analysis and Problem Complexity
  
• 出版者：Springer Berlin / Heidelberg
• ISSN：1611-3349

文摘

The nearest neighbor approach is one of the most successfully deployed techniques used for sensor-based activity recognition. Nevertheless, this approach presents some disadvantages in relation to response time, noise sensitivity and high storage requirements. The response time and storage requirements are closely related to the data size. This notion of data size is an important issue in sensor-based activity recognition given the vast amounts of data produced within smart environments. A wide range of prototype generation algorithms, which are designed for use with the nearest neighbor approach, have been proposed in the literature to reduce the size of the data set. These algorithms build new artificial prototypes, which represent the data, and subsequently lead to an increase in the accuracy of the nearest neighbor approach. In this work, we discuss the use of prototype generation algorithms and their effect on sensor-based activity recognition using the nearest neighbor approach to classify activities, reducing the response time. A range of prototype generation algorithms based on positioning adjustment, which reduce the data size, are evaluated in terms of accuracy and reduction. These approaches have been compared with the normal nearest neighbor approach, achieving similar accuracy and reducing the data size. Analysis of the results attained provide the basis for the use of prototype generation algorithms for sensor-based activity recognition to reduce the overall response time of the nearest neighbor approach.