TALK: Time-Varying Mixing Models for ICA Extraction
new algorithms for blind extraction of independent vectors
Friday, October 25, 2019 · 3 - 4 PM
Time-Varying Mixing Models for Independent Component/Vector Extraction
Prof. Zbyněk Koldovský, Technical University of Liberec, Czech Republic
3:00pm Friday, 25 October 2019, ITE 217b, UMBC
In this talk, we will go back to the fundamental model used in independent component and vector analysis. It is not the statistical model embodying the signals' independence but the instantaneous linear mixing model describing the mechanism how sources are mixed together. We will introduce new mixing models that are suitable for describing piecewise mixing processes, involving time-varying mixtures. Then, using the standard statistical model and maximum likelihood principle, we will derive new algorithms for the blind extraction of independent (vector) sources.
Zbyněk Koldovský received the M.S. degree and Ph.D. degree in mathematical modeling from Faculty of Nuclear Sciences and Physical Engineering at the Czech Technical University in Prague in 2002 and 2006, respectively. He is currently an associate professor at the Institute of Information Technology and Electronics, the leader of Acoustic Signal Analysis and Processing Group, and the Vice-dean for Science, Research and Doctoral Studies at Faculty of Mechatronics, Informatics and Interdisciplinary Studies, Technical University of Liberec. He has also been with the Institute of Information Theory and Automation of the Academy of Sciences of the Czech Republic since 2002. His main research interests are focused on audio signal processing, blind source separation, statistical signal processing and multilinear algebra.