Keynote speakers

Dr. Karen Egiazarian (Eguiazarian)
Tampere University, Finland

Digital Image Restoration: Past, Present and Future

Abstract: This talk is about digital image restoration, with the aim at taking a corrupted (e.g. noisy, blurred) image and estimating the clean, original image. It is one of the most important fields in digital image processing. We give a short overview of image restoration methods, analyse its current state and potential future trends. Specifically, methods utilizing sparsity and self-similarity image priors, as well as learning-based methods, such as those based on convolutional neural networks, will be covered.

Karen O. Egiazarian (Eguiazarian) (Fellow, IEEE, 2018) received M.Sc. in mathematics from Yerevan State University, Armenia, in 1981, the Ph.D. degree in physics and mathematics from Moscow State University, Russia, in 1986, and a Doctor of Technology from Tampere University of Technology, Finland, in 1994. In 2015 he has received the Honorary Doctoral degree from Don State-Technical University (Rostov-Don, Russia). Dr. Egiazarian is a co-founder and CEO of Noiseless Imaging Oy (Ltd), Tampere University of Technology spin-off company. He is a Professor at Computational Sciences unit, Tampere University, Tampere, Finland, leading ‘Computational imaging’ group, and Docent in the Department of Information Technology, University of Jyväskyla, Finland. Dr. Egiazarian has published over 700 refereed journal and conference articles, books and patents in these fields. He is a recipient of IS&T Service Award in 2014, is Editor-in-Chief of Journal of Electronic Imaging (SPIE), and member of the DSP Technical Committee of the IEEE Circuits and Systems Society.

Professor Veljko Milutinovic
Adjunct Professor of Computer Science at Indiana University in Bloomington, IND, USA, Life Member of the ACM, Life Fellow of the IEEE, Member, a Former Trustee and Treasurer, of Academia Europaea, Founding Member of the Serbian National Academy of Engineering, Foreign Member of the Montenegro National Academy of Sciences and Arts

Data Flow Super Computing for Big Data Deep Analytics and Signal Processing

Abstract: Acording to Alibaba and Google, as well as the open literature, the DataFlow paradigm, compared to the ControlFlow paradigm, offers: (a) Speedups of at least 10x to 100x and sometimes much more (depends on the algorithmic characteristics of the most essential loops and the spatial/temporal characteristics of the Big Data Streem, etc.), (b) Potentials for a better precision (depends on the characteristics of the optimizing compiler and the operating system, etc.), (c) Power reduction of at least 10x (depends on the clock speed and the internal architecture, etc.), and (d) Size reduction of well over 10x (depends on the chip implementation and the packiging technology, etc.). However, the programming paradigm is different, and has to be mastered. This 30-minute lecture, followed by a 90-minute tutorial on DataFlow programming, analyses the essence of DataFlow SuperComputing, defines its advantages and sheds light on the related programming model that corresponds to the recent Intel patent about the future Intel's dataflow processor. The stress is on issues of interest for Signal Processing.

Prof. Veljko Milutinovic (1951) received his PhD from the University of Belgrade in Serbia, spent about a decade on various faculty positions in the USA (mostly at Purdue University and more recenlty at the Indiana University in Bloomington), and was a co-designer of the DARPAs first GaAs RISC microprocessor at 200MHz (about a decade before commercial efforts on the same speed) and the DARPAs first GaAs Systolic Array with 4096 processors on 200MHz (both well documented in the open literature). Later, for about three decades, he taught and conducted research at the University of Belgrade, in EE, MATH, BA, and PHYS/CHEM. Now he serves as the Chairman of the Board of IPSI Belgrade (a spin-off of Fraunhofer IPSI from Darmstadt, Germany). His research is mostly in datamining algorithms and dataflow computing, with the emphasis on mapping of data analytics algorithms onto fast energy efficient architectures. For 18 of his books and related publications, forewords were written by 18 different Nobel Laureates with whom he cooperated on his past industry sponsored projects. He has over 100 SCI journal papers (mostly in IEEE and ACM journals), well over 1000 Thomson-Reuters citations, well over 1000 SCOPUS citations and well over 4000 Google Scholar citations, with h=36 and i10=100. Short or long courses on the subject he delivered so far in a number of universities worldwide.

Professor Victor Fedorov
Pirogov Medical University, RUDN University, MIPT, Russia

Heart rate variations analysis: traditions, misconceptions, perspectives

Abstract: The article is devoted to the heart rate "variability" analysis from the history of the issue to the current Euro-American and Russian recommendations. The aim of this work was to analyze the correctness of these documents from the point of view of both physiology and algorithms for identifying and applying quantitative parameters of heart rate variations. The methodological and algorithmic imperfection of the applied approaches is shown. Ways to overcome the existing problems are proposed.

Doctor of medical sciences Victor Fedorovich Fedorov is a professor of Pirogov Russian National Research Medical University, Biomedical faculty, (Moscow); a professor of RUDN-University, Medical Institute (Moscow); a professor of Phystech School of Radio Engineering and Computer Technology of Moscow Institute of Physics and Technology (National Research University) (Dolgoprudny). His research interests are: physiological cybernetics, functional diagnostics, telemedicine, medical instrumentation, signal processing. Professor Fedorov is an author of more than 100 scientific publication and 11 patents and software registration certificates.


Ph.D. Maxim Vashkevich

Belarussian State University of Informatics and Radioelectronics (BSUIR) 

Psychoacoustically motivated and bioinspired methods of
digital signal processing

Abstract: The talk is dedicated to the methods of digital signal processing used in the models of auditory perception. Such models are widely used in audio coding, speech recognition, speech enhancement, noise reduction, etc. A brief overview of approaches to design a models of auditory filters is given. Attention is paid to the problem of implementation of auditory filters. A system for voice pathology detection based on the analysis of the modulation spectrum of speech in critical bands is considered as a new promising direction for the application of models of auditory perception.

Vashkevich Maxim graduated from BSUIR in 2008 with a degree in computer engineering, and in 2013 he defended his Ph.D. thesis. In 2014 a monograph was published (together with Petrovsky A.A. and Azarov I.S.): “Cosine-modulated filter banks with all-pass transform in hearing aid design”. Since 2012 he has been working as an associate professor at the Computer Engineering Department of BSUIR. He has over 80 scientific papers. His main research interests lie in the areas of digital processing of speech signals, the theory of filter banks, methods of signal processing in hearing aids and methods of detecting voice pathologies detection.


The list of plenary papers will be expanded further...