Plenary Speaker: Professor Visa Koivunen

Aalto Distinguished Professor, D.Sc.(EE)

Department of Information and Communications Engineering, Aalto University, Finland

Lecture title: Machine Learning for Radio Frequencies

Abstract

Machine learning (ML) has had tremendous success stories in perception systems such as speech recognition and image analysis.  It has provided new capabilities for extracting relevant information from physical world. In this talk, topics on modern data and model driven machine learning and related signal processing for Radio Frequency (RF) data are discussed. ML methods allow for improving the performance of modern RF systems, including fully adaptive cognitive radars, 5G and beyond wireless communications, security, well-being and imaging applications. ML methods are particularly suitable for problems and applications that may be subject to modeling or algorithmic deficits because of the lack of knowledge of the underlying physical phenomena, complexity or dimensionality of the problem. We will also discuss pros and cons of ML and broader General AI in this context. We will provide insight on learning methods specific to RF spectrum with application examples on supervised learning using complex-valued IQ-data and complex deep neural networks based on Wirtinger calculus for RF waveform classification, reinforcement learning in agile spectrum use and resource allocation in Integrated Sensing and Communications (ISAC) and multifunction radar systems, MIMO radar waveform synthesis using Generative Adversarial Networks (GAN), and privacy preserving distributed learning. We will also consider building situational awareness about electromagnetic spectrum that is needed for optimizing and learning in adaptive RF systems.

Biography

Visa Koivunen (IEEE Fellow, EURASIP Fellow) received his D.Sc. (EE) degree with honors from the Univ of Oulu, Dept. of Electrical Engineering. He was a visiting researcher at the Univ of Pennsylvania, Philadelphia, USA, 1991-1995 and adjunct full professor in 2003-2006. Since 1999 he has been a full Professor of Signal Processing at Aalto University (formerly Helsinki UT), Finland. He received the Academy professor position in 2010 and Aalto Distinguished professor in 2020. During his sabbatical terms in 2006-2007 and 2013-2014 he was a visiting faculty at Princeton University. He has also been a Visiting Fellow at Nokia Research (2006-2012). Since 2010 he has spent mini-sabbaticals at Princeton University each year. On his sabbatical term in 2022-23, he was visiting professor at Adaptive Systems Laboratory, EPFL, Lausanne, Switzerland.

Dr. Koivunen's research interest include statistical signal processing, wireless comms, radar, multisensor systems and machine learning. He was awarded the IEEE SP Society best paper award for the year 2007 (with J. Eriksson) and 2017 (w Zoubir, Muma and Chakhchouk) and received multiple best conference paper awards . He has served in editorial board for the Proceedings of the IEEE, IEEE SP Letters, IEEE TR on SP, and IEEE SP Magazine. He has served in IEEE Fourier Award, Kilby medal and Fellow Evaluation committees and SPS Award Board. He was awarded the 2015 EURASIP (European Association for Signal Processing) Technical Achievement Award for fundamental contributions to statistical signal processing and its applications in wireless communications, radar and related fields. 

Plenary Speaker: Professor Luis Cruz

Department of Electrical and Computer Engineering, University of Coimbra, Portugal 

Lecture title: An overview of the JPEG Point Cloud Coding Standardization Activities 

Abstract

The increasing use of immersive visual data based on representations like light field image, digital holographic signals, meshes and point clouds led to the creation of the ISO/IEC JTC 1/SC 29 JPEG Pleno project aimed at developing standards for the efficient coding of these type of volumetric data. The project is structured as three distinct activities focused on light field image coding, holographic signals coding and point cloud coding. This talk will present an overview of the latter activity, from the early explorations on possible point cloud coding solutions, quality evaluation methods and test sets, to the design and development of a coding solution based on deep-learning principles. The presenter will finish with a reflection on the challenges still to be overcome until the proposed codec becomes an ISO/IEC standard.

Biography

Luis Cruz received the Licenciado and M.Sc. degrees in Electrical Engineering from the University of Coimbra, Portugal, in 1989 and 1993 respectively. He also holds an MSc degree in Mathematics and a Ph.D. degree in Electrical Computer and Systems Engineering from Rensselaer Polytechnic Institute (RPI), Troy, NY, US granted in 1997 and 2000 respectively. He has been with the Department of Electrical and Computer Engineering of the University of Coimbra in Portugal since 1990 first as a Teaching Assistant and as an Assistant Professor since 2000.

He is a senior researcher of the Instituto de Telecomunicações (Coimbra) where he works on image and video processing and coding and (bio)medical image processing. Currently he co-chairs the JPEG Point Cloud Coding Ad-Hoc-Group. He is an IEEE senior member and an EURASIP member.