Predavanje "Entropies: It's Time to Get Rid of Unnecessary Parameters and Computational Intensive Algorithms"

Synopsis: Entropies, and especially Approximate and Sample Entropy, are very often used for the estimation of the complexity of physiological systems, with Heart Rate Variability being one the most common application fields. However, the computation of both Approximate and Sample Entropy has two significant weak points: (a) it is computationally intensive and (b) requires two parameters the value of which is an ad hoc decision. In this presentation we will discuss algorithms which reduce significantly the computational time for the calculation of both Approximate and Sample Entropy. This algorithm excludes comparisons for the identification of similar points for which we can decide very early that they will fail. We will also discuss some ideas of how we can define entropy so that we can express the complexity of a system without be necessary to define the two parameters. Experimental results from the field of Heart Rate Variability analysis will be presented, using a database with recordings from congestive heart failure patients and one with healthy subjects.

George Manis is an Assistant Professor in the Department of Computer Science & Engineering, University of Ioannina, Greece. He received his first Degree (5 year Diploma Degree) from the School of Electrical and Computer Engineering, National Technical University of Athens, Greece, in 1992, his MSc in Parallel and Distributed Systems from Queen Mary, University of London, UK, in 1993 and his PhD again from the School of Electrical and Computer Engineering, National Technical University of Athens, in 1997. Since the year 2000, he serves at the Department of Computer Science & Engineering in the University of Ioannina, currently as an Assistant Professor. His research interests include Computer Systems and Biomedical Engineering. He has published more than 80 research papers in scientific journals and conferences.