EUSIPCO 2019 logo

Keynote Speakers

KeyNote Speakers

Josiane Zerubia

INRIA Sophia Antipolis Méditerranée, France.

September, 3rd 9:30-10:30

Marked point process models in image processing. Application to Remote Sensing.

Josiane Zerubia

Josiane Zerubia has been a permanent research scientist at INRIA since 1989 and director of research since July 1995. She was successively head of 3 laboratories in remote sensing from 1995 to 2016. She has been adjunct-professor at ISAE-SUPAERO in Toulouse since 1999. Her main research interest is in signal and image processing using probabilistic models. She also works on parameter estimation, statistical learning and optimization techniques. In terms of applications, she worked on speech processing (1982-1988), biological image processing (2001-2011), skin imaging (2009-2018) and remote sensing (1988-). She published a book on Markov random fields in image segmentation in 2012 (Now pub.), co-authored with Prof. Zoltan Kato. She was co-editor with Prof. Gabriele Moser of a book on mathematical models for remote sensing image processing in 2018 (Springer pub.). She has been nominated EURASIP Fellow in 2019. She is also a Fellow of the IEEE (2003) and was IEEE SP Society Distinguished Lecturer (2016-2017). She received the excellency award from Université Cote d'Azur (UCA) in 2016, several best paper awards with her students and collaborators, and was made "Chevalier de l'Ordre National du Mérite" by the President of the French Republic in 2002 for an exemplary career in research.

Date: September, 3rd 9:30-10:30

Title: Marked point process models in image processing. Application to Remote Sensing.

Abstract:Stochastic methods are now widespread in image analysis. They have proved to be powerful tools to solve inverse problems such as image classification or restoration. Since the mid-nineties, many research works have extended the initial pixel based approach to the concept of object in order to deal with shape detection problems. In particular, stochastic models have shown good potentialities in extracting simple shapes. Generally, configurations of parametric functions are sampled from probability distributions defined in a configuration space, Markov Chain Monte Carlo (MCMC) being one of the most popular families of samplers. In various application domains, from line detection to 3D reconstruction, the MCMC samplers are efficient for object extraction in large configuration spaces from any type of probability distribution. Models based on marked point processes are among the most efficient stochastic approaches and have lead to convincing experimental results in various shape detection applications (such as extraction of line segments, rectangles, circles, ellipses, …). The marked point processes exploit random variables whose realizations are configurations of geometrical objects. After specifying a probability distribution measuring the quality of each object configuration, the maximum density estimator is searched for by MCMC techniques coupled with a stochastic relaxation. Such processes are especially adapted to the description of complex spatial interactions between the objects. Various examples on road network detection, crown tree extraction, flamingo counting; boat detection or building reconstruction will be given during the talk. Then, an extension of this framework based on spatio-temporal marked point process models to jointly detect and track multiple objects in image sequences will be described. Finally, results on both on high-resolution satellite and drone image sequences will be presented.

Yurii Nesterov

Center for Operations Research and Econometrics (CORE), Université Catholique de Louvain (UCL), Belgium.

September, 4th 9:00-10:00

Relative smoothness: new paradigm in Convex Optimization.

Yurii Nesterov

Yurii Nesterov is a professor at the Center for Operations Research and Econometrics (CORE) in the Catholic University of Louvain (UCL), Belgium. He received Ph.D. degree (Applied Mathematics) in 1984 at Institute of Control Sciences, Moscow. Starting from 1993 he works at Center of Operations Research and Econometrics (Catholic University of Louvain, Belgium). His research interests are related to complexity issues and efficient methods for solving various optimization problems. The main results are obtained in Convex Optimization (optimal methods for smooth problems, polynomial-time interior-point methods, smoothing technique for structural optimization, complexity theory for second-order methods, optimization methods for huge-scale problems). He is an author of 5 monographs and more than 100 refereed papers in the leading optimization journals. He got several international prizes, among which there are Dantzig Prize from SIAM and Mathematical Programming society (2000), von Neumann Theory Prize from INFORMS (2009), SIAM Outstanding paper award (2014), and Euro Gold Medal from Association of European Operations Research Societies (2016). In 2018 he won an Advanced Grant from the European Research Council.

Date: September, 4th 9:00-10:00

Title: Relative smoothness: new paradigm in Convex Optimization.

Abstract: Relative Development and computational abilities of optimization methods crucially depend on the auxiliary tools provided to them by the method's designers. During the first decades of Convex Optimization, the methods were based either on the proximal setup, allowing Euclidean projections onto the basic feasible sets, or on the linear minimization framework, which assumes a possibility to minimize a linear function over the feasible set. However, recently it was realized that any possibility of simple minimization of an auxiliary convex function leads to the efficient minimization methods for some family of more general convex functions, which are compatible with the first one. This compatibility condition, called relative smoothness, was firstly exploited for smooth convex functions (Bauschke, Bolt and Teboulle, 2016) and smooth strongly convex functions (Lu, Freund and Nesterov, 2018). In this talk we make the final step and show how to extend this framework onto the class of nonsmooth functions. We also discuss possible consequences and applications.

Raymond Knopp

EURECOM, Sophia Antipolis, France.

September, 5th 9:30-10:30

Academic research, Standardization and Open-Prototyping

Raymond Knopp

Raymond Knopp is professor in the Communication Systems Department at EURECOM. He is also currently a part-time visiting professor at the Beijing University of Posts and Telecommunications under the Discipline Innovative Engineering Plan. His current research and teaching interests are in the area of digital communications, software radio architectures, and implementation aspects of signal processing systems and real-time wireless networking protocols. He has a proven track record in managing both fundamental and experimental research projects at an international level and is also General Secretary of the open-source academia-industry radio platform initiative which aims to bridge the gap between cutting-edge theoretical advances in wireless communications and practical designs.

Date: September, 5th 9:30-10:30

Title: Academic research, Standardization and Open-Prototyping

Abstract: This talk provides an overview of EURECOM's vision regarding direct  contributions of innovations from its research teams to standardization bodies such as the 3GPP, in particular related to physical layer processing. We then discuss the relationship with our work on using open source code and community-based development through the OpenAirInterface Software Alliance to prototype both standardized and pre-standardized innovations. In particular, we will shed some light on how such methodologies can coexist effectively with industrial standardization processes and more importantly how academic institutions can hope to provide more direct impact to future cellular system architecture.

Alejandro Ribeiro

Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, USA

September, 6th 9:00-10:00

Graph Neural Networks

Alejandro Ribeiro

Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, USA

Alejandro Ribeiro received the B.Sc. degree in electrical engineering from the Universidad de la Republica Oriental del Uruguay, Montevideo, in 1998 and the M.Sc. and Ph.D. degree in electrical engineering from the Department of Electrical and Computer Engineering, the University of Minnesota, Minneapolis in 2005 and 2007. From 1998 to 2003, he was a member of the technical staff at Bellsouth Montevideo. After his M.Sc. and Ph.D studies, in 2008 he joined the University of Pennsylvania (Penn), Philadelphia, where he is currently Professor of Electrical and Systems Engineering. His research interests are in the applications of statistical signal processing to collaborative intelligent systems. His specific interests are in wireless autonomous networks, machine learning on network data and distributed collaborative learning. Papers coauthored by Dr. Ribeiro received the 2014 O. Hugo Schuck best paper award, and paper awards at CDC 2017, SSP Workshop 2016, SAM Workshop 2016, Asilomar SSC Conference 2015, ACC 2013, ICASSP 2006, and ICASSP 2005. His teaching has been recognized with the 2017 Lindback award for distinguished teaching and the 2012 S. Reid Warren, Jr. Award presented by Penn's undergraduate student body for outstanding teaching. Dr. Ribeiro is a Fulbright scholar class of 2003 and a PennFellow class of 2015.

Date: September, 6th 9:00-10:00

Title: Graph Neural Networks

Abstract: Convolutional Neural Networks (CNN) are layered information processing architectures in which each of the layers is itself the composition of a convolution operation with a pointwise nonlinearity. Graph Neural Networks (GNNs) replace the regular convolution operation with a graph convolution operation. We will discuss graph convolutions, their use in building GNN architectures, and explore stability properties of GNN operators. The stability results establish that a GNN is stable to graph deformations that are close to permutations. This result provides a theoretical basis to characterize classes of machine learning problems in which we expect GNNs to work well. We will discuss applications to control of large scale collaborative autonomous systems and wireless networks.