Application of Multivariate Empirical Mode Decomposition in EEG Signals for Subject Independent Affective States Classification

Prof. Konstantinos N. Plataniotis
Department of Electrical and Computer Engineering
University of Toronto
CANADA
E-mail: kostas@ece.utoronto.ca
 

Abstract: Physiological signals, EEG in particular, are inherently noisy and non-linear in nature which are challenging to work with using conventional linear signal processing methods. In this paper, we are adopting a new signal processing method, Multivariate Empirical Mode Decomposition, as a preprocessing method to reconstruct EEG signals according to its instantaneous frequencies. To test its effectiveness, we applied this signal reconstruction technique to analyze EEG signals for a 2-dimensional affect states classification application. To evaluated the proposed EEG signal processing system, a three-class classification experiment were carried out on the �Emobrain� dataset from eNTERFACE'06 with K-nearest neighbors (KNN) and Linear Discriminate Analysis (LDA) as classifiers. A leave-one-subject out cross validation process were used and an averaged correct classification rate of 90.77% were achieved. Another main contribution of this paper was inspired by the growth of non-medical grade EEG headsets and its potential in advanced human-computer interface design. However, to reduce cost and invasiveness, consumer grade EEG headsets have far less number of electrodes. In this paper, we used emotion recognition as a case study, and applied Genetic Algorithm to systematically select the critical channels (or sensor locations) for this application. The results of this study will shed lights on the sensor configuration challenges faced by most consumer-grade EEG headset design projects.

Short biography: Konstantinos N. (Kostas) Plataniotis received his B. Eng. degree in Computer Engineering from University of Patras, Greece and his M.S. and Ph.D. degrees in Electrical Engineering from Florida Institute of Technology Melbourne, Florida. He was with the Computer Science Department at Ryerson University, Ontario, Canada from July 1997 to June 1999. Dr. Plataniotis is currently a Professor with The Edward S. Rogers Sr. Department of Electrical and Computer Engineering at the University of Toronto in Toronto, Ontario, Canada, where he directs the Multimedia Laboratory. He is a founding member and the inaugural Director � Research of the Identity, Privacy and Security Institute, IPSI, (www.ipsi.utoronto.ca). Kostas was the Director (January 2010- June 2012) of the Knowledge Media Design Institute, KMDI, (www.kmdi.utoronto.ca ) at the University of Toronto.
Dr. Plataniotis was the Guest Editor for the March 2005 IEEE Signal Processing Magazine special issue on �Surveillance Networks and Services�, and the Guest Editor for the EURASIP Applied Signal Processing Journal�s special issue on �Advanced Signal Processing & Pattern Recognition Methods for Biometrics�. He is a member of the IEEE Periodicals Review and Advisory Committee (2011-2013); he has served as a member of the 2008 IEEE Educational Activities Board; he chaired of the IEEE EAB Continuing Professional Education Committee, and he served as the 2008 representative of the Computational Intelligence Society to the IEEE Biometrics Council. Dr. Plataniotis chaired the 2009 Examination Committee for the IEEE Certified Biometrics Professional (CBP) Program (www.ieeebiometricscertification.org) and he served on the Nominations Committee for the IEEE Council on Biometrics. He was a member of the Steering Committee for the IEEE Transaction on Mobile Computing, an Associate Editor for the IEEE Signal Processing Letters as well as the IEEE Transactions on Neural Networks and Adaptive Systems and he has served as the Editor-in-Chief for the IEEE Signal Processing Letters from January 1, 2009 to December 31, 2011. Dr. Plataniotis chaired the IEEE Toronto Signal Processing and Applications Toronto Chapter from 2000 to 2002, he was the 2004-05 Chair of the IEEE Toronto Section and a member of the 2006 as well as 2007 IEEE Admissions & Advancement Committees. He served as the Technical Program Committee Co-Chair for the 2013 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2013) and he is the Vice President � Membership for the IEEE Signal Processing Society (2014-2016).
Dr. Plataniotis is a Fellow of IEEE, Fellow of the Engineering Institute of Canada, a registered professional engineer in the province of Ontario, and a member of the Technical Chamber of Greece.
The recipient of numerous grants and research contracts as the principal investigator, he speaks internationally and writes extensively in his field and he has been a consultant to a number of companies. He has served as lecturer in 12 short courses to industry and continuing education programs; he is a contributor to seventeen books, the co-author of �Color Image Processing and Applications�, Springer Verlag, 2000, (ISBN-3-540-66953-1) and �WLAN Positioning Systems: Principles & applications in Location-based Services�, Cambridge University Press, 2012 (ISBN 978-0-521-9185-2), �Multi-linear Subspace Learning: Reduction of multi-dimensional data}, CRC Press, 2013, (ISBN: 978-14398557243). He is the co-editor of �Color Imaging: Methods and Applications�, CRC Press, September 2006, (ISBN 084939774X) and the Guest Editor of the IEEE/Wiley Press volume on �Biometrics: Theory, Methods and Applications� published in October 2009 (ISBN: 9780470247822). Dr. Plataniotis has published more than 400 papers in refereed journals and conference proceedings. In 2005 he became the recipient of the IEEE Canada Engineering Educator Award for �contributions to engineering education and inspirational guidance of graduate students�. Dr. Plataniotis is the joint recipient of the �2006 IEEE Trans. on Neural Networks Outstanding Paper Award� for the published in 2003 �Face recognition using kernel direct discriminant analysis algorithms�, IEEE Trans. on Neural Networks, Vol. 14, No 1, 2003..