Traffic-Aware Scheduling and Feedback Reporting in Wireless Networks

Matha DEGHEL
Soutenance de thèse de doctorat le 22 Mai 2017, 09h00 à CentraleSupelec (Gif-sur-Yvette) Amphi Mesny

Demand of wireless communication systems for high throughputs continues to increase, and there are no signs this trend is slowing down. Three of the most prominent techniques that have emerged to meet such demands are OFDMA, cooperative relaying and MIMO. To fully utilize the capabilities of systems applying such techniques, it is essential to develop efficient scheduling algorithms and, more generally, efficient resource allocation algorithms. Classical studies on this subject investigate in much detail settings where the data requests of the users are not taken into consideration or where the perfect and full CSI is assumed to be available for the scheduling mechanism. In practice, however, different limitations may result in not having perfect or full CSI knowledge, such as limited feedback resources, probing cost and delay in the feedback process. Accordingly, in this thesis we examine the problems of scheduling and feedback allocations under realistic considerations concerning the CSI knowledge. Analysis is performed at the packet level and considers the queueing dynamics in the systems with arbitrary arrival processes, where the main performance metric we adopt is the stability of the queues. The first part of the thesis considers a multipoint to multipoint MIMO system with TDD mode under limited backhaul capacity and taking into account the feedback probing cost. Regarding the interference management technique, we apply interference alignment (IA) if more than one pair are active and SVD if only one pair is active. The second part of the thesis considers a multiuser multichannel OFDMA-like system where delayed and limited feedback is accounted for. Two scenarios are investigated, namely the system without relaying and the system with relaying. For the latter one, an additional imperfection we account for is that the users have incomplete knowledge of the fading coefficients between the base-station and the relay.

Jury:

M. Vincent K. N. LAU  Hong Kong       University of Science and Technology  Rapporteur

M. Jean-Marie GORCE                        NSA Lyon                                             Rapporteur

M. Anthony EPHREMIDES                  University of Maryland, College Park   Examinateur

M. Jianwei  HUANG Chinese               University of Hong Kong                      Examinateur

M. Michel  KIEFFER                             Université Paris-Sud                           Examinateur

Mme Laura  COTTATELLUCCI            Eurecom                                              Examinateur

M. Mérouane DEBBAH                        CentraleSupélec                                 Directeur de thèse

M. Mohamad ASSAAD                        CentraleSupélec                                 CoDirecteur de thèse

S³ seminar : Deux trous noirs dans une meule de foin : analyse de données pour l'astronomie gravitationnelle

Séminaire le 19 Mai 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle des séminaires du L2S
Eric Chassande-Mottin, (CNRS, AstroParticule et Cosmologie, Université Paris Diderot)


Le 14 septembre 2015, les deux détecteurs du Laser Interferometer Gravitational-wave Observatory (LIGO) inauguraient une nouvelle ère pour l'astrophysique en observant pour la première fois une onde gravitationnelle issue de la fusion de deux trous noirs faisant chacun trente fois la masse du soleil environ et situés à une distance supérieure à un milliard d'années-lumière. On donnera une vue d'ensemble de cette découverte majeure en insistant sur les méthodes d'analyse de données utilisées pour sortir le signal du bruit complexe rencontré dans ces expériences.

Coded Caching in Wireless Networks

Séminaire le 19 Mai 2017, 09h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Meixia Tao


The global mobile data traffic has been shifting from voice and messages to rich content distributions, such as video streaming and application downloads. These contents are typically produced ahead of transmission and can be requested by multiple users though at possibly different times. By prefetching popular contents during off-peak times at the edge of wireless networks, such as small base stations, helper nodes, and user devices, wireless caching can alleviate peak-time network congestion and reduce user access latency. A fundamental question is what and how much gain can be leveraged by caching. In this talk, we shall investigate the gain of caching in two types of wireless networks. One is a general wireless interference network with arbitrary number of transmitters and arbitrary number of receives and with caches equipped at all the nodes. An information-theoretic study in terms of the storage-latency tradeoff will be presented. The other is a large-scale small-cell network where each small base station is equipped with a cache. We apply stochastic geometry to model, analyze, and optimize coded caching with performances characterized by average fractional offloaded traffic and average ergodic rate. Our study reveals several design insights of caching in practical wireless networks.

Bio: Meixia Tao received the B.S. degree from Fudan University, Shanghai, China, in 1999, and the Ph.D. degree from Hong Kong University of Science and Technology in 2003. She is currently a Professor with the Department of Electronic Engineering, Shanghai Jiao Tong University, China. Prior to that, she was a Member of Professional Staff at Hong Kong Applied Science and Technology Research Institute during 2003-2004, and a Teaching Fellow then an Assistant Professor at the Department of Electrical and Computer Engineering, National University of Singapore from 2004 to 2007. Her current research interests include content-centric wireless networks, wireless caching and multicasting, resource allocation, and interference coordination.

Dr. Tao is currently serving as a member of the Executive Editorial Committee of the IEEE Transactions on Wireless Communications and an Editor for the IEEE Transactions on Communications. Dr. Tao is the recipient of the IEEE Heinrich Hertz Award for Best Communications Letters in 2013 and the IEEE ComSoc Asia-Pacific Outstanding Young Researcher Award in 2009. She also receives the best paper awards from IEEE/CIC ICCC 2015 and IEEE WCSP 2012.

Semantically-Secured Message-Key Trade-off over Wiretap Channels with Random Parameters

Séminaire le 4 Mai 2017, 14h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Alexander Bunin, Faculty of Electrical Engineering at the Technion-Israel Institute of Technology, Israel.


Modern communication systems usually present an architectural separation between error correction and data encryption. The former is typically realized at the physical layer by transforming the noisy communication channel into a reliable "bit pipe". The data encryption is implemented on top of that by applying cryptographic principles.

Physical layer security (PLS) rooted in information-theoretic principles, is an alternative approach to provably secure communication. This approach dates back to Wyner's celebrated 1975 paper on the wiretap channel (WTC). By harnessing randomness from the noisy communication channel and combining it with proper physical layer coding, PLS guarantees protection against computationally-unlimited eavesdroppers with no requirement that the legitimate parties share a secret key (SK) in advance. The eavesdroppers computational abilities are of no consequence here since the signal he/she observes from the channel carries only negligible information about the secret data.

Two fundamental questions in PLS are those of the best achievable transmission rate of a secret message (SM) over a noisy channel, and the highest attainable SK rate that distributed parties can agree upon. Our work studies the trade-off between SM and SK rates simultaneously achievable over a state-dependent wiretap channel WTC with non-causal channel state information (CSI) at the encoder. This model subsumes all other instances of CSI availability as special cases, and calls for an efficient utilization of the state sequence both for reliability and security purposes. We derive an inner bound on the SM-SK capacity region based on a novel superposition coding scheme. Our inner bound improves upon the previously best known SM-SK trade-off result by Prabhakaran et al., and to the best of our knowledge, upon all other existing lower bounds for either SM or SK for this setup. The results are derived under the strict semantic-security metric that requires negligible information leakage for all message-key distributions. The achievability proof uses the strong soft-covering lemma for superposition codes.
The talk is based on joint work with: Z. Goldfeld and H. H. Permuter (Ben-Gurion University of the Negev, Israel), S. Shamai (Technion - Israel Institute of Technology), P. W. Cuff (Princeton University) and P. Piantanida (CentralSupelec).
The work of A. Bunin and S. Shamai has been supported by the European Union's Horizon 2020 Research And Innovation Programme, grant agreement no. 694630.

Séminaire d’Automatique du plateau de Saclay : Chance-constrained optimization with tight confidence bounds.

Séminaire le 25 Avril 2017, 11h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Mark Cannon (Department of Engineering Science, University of Oxford)


Optimization problems involving chance constraints, i.e. constraints on decision variables that are required to hold with given probability, occur in numerous engineering contexts, but have in particular contributed to developments in robust control design over the past decade. This talk will provide a new perspective on randomized methods for solving chance-constrained programming problems based on samples of uncertain parameters, and presents a theoretical framework for sampled convex programming that encompasses analyses of constraint discarding approaches and sequential sampling approaches. We show that tight bounds on the confidence of the solution of a sampled problem meeting chance constraints apply if a randomized sample discarding strategy is employed. This suggests a solution methodology which is both computationally convenient, being based on the solutions of convex sampled optimization problems, and efficient, requiring the solution of smaller numbers of problems than existing constraint discarding and sequential methods. We illustrate the method with examples from stochastic model predictive control design.

Bio. Mark Cannon obtained the degrees of MEng in Engineering Science in 1993 and D.Phil. in Control Engineering in 1998, both from Oxford University, and SM in Mechanical Engineering in 1995 from Massachusetts Institute of Technology. He is currently Associate Professor of Engineering Science, Oxford University, and an Official Fellow of St John's College. His research interests are in robust constrained control and stochastic model predictive control, specifically: issues relating to optimization and controller design, closed loop stability and constraint satisfaction, and applications to power management in electric vehicles.

Séminaire d’Automatique du plateau de Saclay : On generalized homogeneity and time-constrained stabilization of evolution systems.

Séminaire le 25 Avril 2017, 10h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Andrey polyakov (NON-A team, Inria-Lille-Nord Europe)


Homogeneity is a symmetry of an object (e.g. function or operator) with respect to some transformations (dilations). Nonlinear homogeneous ODEs (ordinary differential equations) form an important class of models of control systems. They appear as local approximations of nonlinear plant and include models of process control, nonholonomic mechanical systems, models with frictions, etc. Being non-linear the homogeneous systems demonstrate properties typical for linear systems, for example, local stability implies the global one, stable homogeneous control system is ISS (input-to-state stable) with respect to measurement noises and additive exogenous disturbance, etc.

This talk is devoted to extension of ideas of homogeneity to evolution systems in Banach/Hilbert spaces. A lot of well-known partial differential equations are homogeneous in a generalized sense, (e.g. heat, wave, Navier-Stocks, Saint-Venant, Korteweg-de Vries, fast diffusion equations). They inherit many important properties of homogeneous ODEs such as  scalability of trajectories or finite-time stability in the case of negative homogeneity degree. Homogeneity allows us to design a universal control for finite-time stabilization of evolution system.

Bio. Andrey Polyakov received PhD degree from Voronezh State University (Russia) in 2005.
He was lecturer (2004-2007) and associate professor (2008-2010) of this university.
In 2007 and 2008 he was a post-doctoral research associate with Automatic Control Department of CINVESTAV (Mexico). From 2010 till 2013 he was researcher with Institute of Control Sciences of Russian Academy of Sciences. Now Andrey Polyakovis Inria researcher with NON-A team of Inria Lille Nord Europe (France). His research interests include
different problems of robust nonlinear control and estimation. He is co-author of the book "Attractive Ellipsoids in Robust Control". He is editor of International Journal of Robust and Nonlinear Control, Journal of Optimization Theory and Applications (JOTA), Automation and Remote Control.

S³ seminar : Extending Stationarity to Graph Signal Processing: a Model for Stochastic Graph Signals

Séminaire le 31 Mars 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Benjamin Girault, (University of Southern California)


During the past few years, graph signal processing has been extending the field of signal processing on Euclidean spaces to irregular spaces represented by graphs. We have seen successes ranging from the Fourier transform, to wavelets, vertex-frequency (time-frequency) decomposition, sampling theory, uncertainty principle, or convolutive filtering. One missing ingredient though are the tools to study stochastic graph signals for which the randomness introduces its own difficulties. Classical signal processing has introduced a very simple yet very rich class of stochastic signals that is at the core of the study of stochastic signals: the stationary signals. These are the signals statistically invariant through a shift of the origin of time. In this talk, we study two extensions of stationarity to graph signals, one that stems from a new translation operator for graph signals, and another one with a more sensible interpretation on the graph. In the course, we show that attempts of alternate definitions of stationarity on graphs in the recent literature are actually equivalent to our first definition. Finally, we look at a real weather dataset and show empirical evidence of stationarity.

Bio: Benjamin Girault received his License (B.Sc.) and his Master (M.Sc.) in France from École Normale Supérieure de Cachan, France, in 2009 and 2012 respectively in the field of theoretical computer science. He then received his PhD in computer science from École Normale Supérieure de Lyon, France, in December 2015. His dissertation entitled "Signal Processing on Graphs - Contributions to an Emerging Field" focused on extending the classical definition of stationary temporal signals to stationary graph signal. Currently, he is a postdoctoral scholar with Antonio Ortega and Shri Narayanan at the University of Southern California continuing his work on graph signal processing with a focus on applying these tools to understanding human behavior.

S³ seminar : Novel Algorithms for Automated Diagnosis of Neurological and Psychiatric Disorders

Séminaire le 28 Mars 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Hojjat ADELI, (The Ohio State University, Columbus, USA)


Novel algorithms are presented for data mining of time-series data and automated electroencephalogram (EEG)-based diagnosis of neurological and psychiatric disorders based on adroit integration of three different computing technologies and problem solving paradigms: neural networks, wavelets, and the chaos theory. Examples of the research performed by the author and his associates for automated diagnosis of epilepsy, the Alzheimer’s Disease, Attention Deficit Hyperactivity Disorder (ADHD), autism spectrum disorder (ASD), and Parkinson’s disease (PD) are reviewed.

Biography: Hojjat Adeli received his Ph.D. from Stanford University in 1976 at the age of 26. He is Professor of Civil, Environmental, and Geodetic Engineering, and by courtesy Professor of Biomedical Informatics, Biomedical Engineering, Neuroscience, and Neurology at The Ohio State University. He has authored over 550 publications including 15 books. He is the Founder and Editor-in-Chief of international research journals Computer-Aided Civil and Infrastructure, now in 32nd year of publication, and Integrated Computer-Aided Engineering, now in 25th year of publication, and the Editor-in-Chief of International Journal of Neural Systems. In 1998 he received the Distinguished Scholar Award from OSU, “in recognition of extraordinary accomplishment in research and scholarship”. In 2005, he was elected Distinguished Member, ASCE: "for wide-ranging, exceptional, and pioneering contributions to computing in civil engineering and extraordinary leadership in advancing the use of computing and information technologies in many engineering disciplines throughout the world.” In 2010 he was profiled as an Engineering Legend in the ASCE journal of Leadership and Management in Engineering, and Wiley established the Hojjat Adeli Award for Innovation in Computing. In 2011 World Scientific established the Hojjat Adeli Award for Outstanding Contributions in Neural Systems. He is a Fellow of IEEE, the American Association for the Advancement of Science, American Neurological Society, and American Institute for Medical and Biomedical Engineering. Among his numerous awards and honors are a special medal from Polish Neural Network Society, the Eduardo Renato Caianiello Award for Excellence in Scientific Research from the Italian Society of Neural Networks, the Omar Khayyam Research Excellence Award from Scientia Iranica, an Honorary Doctorate from Vilnius Gediminas Technical University, and corresponding member of the Spanish Royal Engineering Society.

S³ seminar : Stochastic proximal algorithms with applications to online image recovery

Séminaire le 24 Mars 2017, 11h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Jean-Christophe PESQUET, CVN, CentraleSupélec


Stochastic approximation techniques have been used in various contexts in machine learning and adaptive filtering. We investigate the asymptotic behavior of a stochastic version of the forward-backward splitting algorithm for finding a zero of the sum of a maximally monotone set-valued operator and a cocoercive operator in a Hilbert space. In our general setting, stochastic approximations of the cocoercive operator and perturbations in the evaluation of the resolvents of the set-valued operator are possible. In addition, relaxations and not necessarily vanishing proximal parameters are allowed. Weak almost sure convergence properties of the iterates are established under mild conditions on the underlying stochastic processes. Leveraging on these results, we propose a stochastic version of a popular primal-dual proximal optimization algorithm, and establish its convergence. We finally show the interest of these results in an online image restoration problem.

Séminaire d'Automatique du plateau de Saclay : Controllability of a bent 3-link magnetic microswimmer

Séminaire le 23 Mars 2017, 11h00 à CentraleSupelec (Gif-sur-Yvette) Salle des séminaires du L2S
Pierre Lissy (CEREMADE, Universite Paris-Dauphine)


A recent promising technique for moving a robotic micro-swimmer (in view notably of medical applications) is to apply an external magnetic field. In this talk, I will focus on a 3-link magnetic microswimmer, which consists of three rigid magnetized segments connected by two torsional springs, one of the springs being twisted, so that the swimmer is not aligned at rest. By acting on it with
an external magnetic field, the swimmer twists and moves through the surrounding fluid. After explaining some specific difficulties coming from the Low Reynolds number regime, I will explain how to model the problem thanks to a system of non-linear ODEs. By considering the external magnetic field as a control function, I will state a local partial controllability result around the equilibrium states. Then, I will propose a constructive method to find a magnetic field that allows the swimmer to move along a prescribed trajectory (tracking) in view of obtaining global partial controllability results. Finally, I will show some numerical simulations that illustrate the practical difficulties of the tracking method due to the straight positions of the swimmer. This is a joint work with Laetitia Giraldi (INRIA Sophia), Jean-Baptiste Pomet (INRIA Sophia) and Clément Moreau (ENS Cachan).

Bio: Pierre Lissy received his Ph.D degree in Applied Mathematics in 2013 under the direction of Professor Jean-Michel Coron at Université Pierre et Marie Curie. He is now Assistant Professor in Applied Mathematics at University Paris-Dauphine since 2014. His research interests include the controllability properties of partial differential equations (PDEs), with a focus on the controllability of (linear and non-linear) coupled systems of PDEs and the estimations of the cost of the control near the minimal time of control and in singular limits. He began recently to work on the controllability properties of magnetized micro-swimmers governed by non-linear ODEs.

Séminaire d'Automatique du plateau de Saclay : Time-extremal trajectories of generic control-affine systems have at most finite-order Fuller singularities

Séminaire le 23 Mars 2017, 10h00 à CentraleSupelec (Gif-sur-Yvette) Salle des séminaires du L2S
Francesco Boarotto (CMAP, Ecole Polytechnique)


Let $M$ be a smooth connected $n$-dimensional manifold, and consider on it the control-affine system $\dot{q}=f_0(q)+uf_1(q),\quad u\in[-1,1].$ Time-extremal trajectories for the time-optimal control problem associated to this system are driven by controls $u$, whose set $\Sigma$ of discontinuities is possibly stratified as follows: $\Sigma_0$ is the set of isolated points in $\Sigma$ (switching times) and, recursively, the $k$-th order Fuller times $\Sigma_k$ are found as the isolated points of $\Sigma\setminus\left(\bigcup_{j=0}^{k-1}\Sigma_j\right)$.

In this talk we show that, in fact, for the generic choice of the pair $(f_0,f_1),$ there exists an integer $N>0$ such that the control $u$ associated to any time-extremal trajectory admits at most Fuller times of order $N$. In particular, $u$ is smooth out of a set of measure zero. This is a joint work with M. Sigalotti (CMAP, Ecole Polytechnique).

Bio: Francesco Boarotto was born in Verona, Italie, in 1988. He received the Master's degree in mathematics from the University of Padou, Italie, in 2012 and the Ph.D degree from SISSA, Trieste, Italie, in 2016. Since then he has been post-doc in CMAP - Ecole Polytechnique. His research interests include geometric control theory and sub-Riemannian geometry.

 

S³ seminar : On Electromagnetic Modeling and Imaging of Defects in Periodic Fibered Laminates

Séminaire le 10 Mars 2017, 12h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Zicheng LIU, (Inverse problems Group, Signals and Statistics Division, L2S Laboratory)


Composite laminates are commonly utilized in industry due to advantages as high stiffness, light weight, versatility, etc. Multiple layers, each one involving periodically-positioned circular-cylindrical fibers in a given homogeneous matrix, are usually involved. However, defects can affect the structure and thereupon impact security and efficiency, and they call for nondestructive testing. By electromagnetic (EM) means, it requires fast and reliable computational modeling of both sound and damaged laminates if one wishes to better understand pluses and minuses of the testing, and derive efficient imaging algorithms for the end user. Both direct modeling and inverse imaging will be introduced in this presentation.  For the former, since the periodicity of the structure is destroyed due to defects, methods based on the Floquet theorem are inapplicable. Two modeling approaches are then utilized: one is with supercell methodology where a fictitious periodic structure is fabricated, so as the EM field solution everywhere in space can be well approximately modeled, provided the supercell be large enough; the other is based on fictitious source superposition (FSS) where defects are treated as equivalent sources and the field solution is a summation of responses to the exterior source and equivalent ones. For imaging, with MUSIC and sparsity-based algorithm, missing fibers could be accurately located.

Biography: Zicheng LIU was born in Puyang, China, in October 1988. He received the M.S. degree in circuit and system from Xidian University, Xi’an, China in March 2014 and is currently pursuing the Ph.D. degree with the benefit of a Chinese Scholarship Council (CSC) grant at the Laboratoire des Signaux et Systèmes, jointly Centre National de la Recherche Scientifique (CNRS), CentraleSupélec, and Université Paris-Sud, Université Paris-Saclay, Paris, France. He will defend his Université Paris-Saclay Ph.D. early Fall 2017. His present work is on the electromagnetic modeling of damaged periodic fiber-based laminates and corresponding imaging algorithms and inversion. His research interests include computational electromagnetics, scattering theory on periodic structures, non-destructive testing, sparsity theory, and array signal processing.

S³ seminar : On Imaging Methods of Material Structures with Different Boundary Conditions

Séminaire le 10 Mars 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Xiuzhu YE, (Beihang University, Beijing, China)


This talk is about the two-dimensional inverse scattering problems for different kinds of boundary conditions. Firstly, we propose a perfect electric conductor (PEC) inverse scattering approach, which is able to reconstruct PEC objects of arbitrary number and shape without requiring prior information on the approximate locations or the number of the unknown scatterers. Secondly, the modeling scheme of the T-matrix method is introduced to solve the challenging problem of reconstructing a mixture of both PEC and dielectric scatterers together. Then the method is further extended to the case of scatterers with four boundary conditions together. Last, we propose a method to solve the dielectric and mixed boundary through-wall imaging problem. Various numerical simulations and experiments are carried out to validate the proposed methods.

Biography: Xiuzhu YE was born in Heilongjiang, China, in December 1986. She received the Bachelor degree of Communication Engineering from Harbin Institute of Technology, China, in July 2008 and the Ph.D. degree from the National University of Singapore, Singapore, in April 2012. From February 2012 to January 2013, she worked in the Department E.C.E., National University of Singapore, as a Research Fellow. Currently, she is Assistant Professor in the School of Electronic and Information Engineering of the Beihang University. She has been and is engaged under various guises with Ecole Centrale de Pékin (ECPK) also. She is presently benefiting from an invited professorship position at University Paris-Sud and later this Summer 2017 she will be benefiting from an invited professorship position at CentraleSupélec, both within the Laboratoire des Signaux et Systèmes, jointly Centre National de la Recherche Scientifique (CNRS), CentraleSupélec, and Université Paris-Sud, Université Paris-Saclay, Gif-sur-Yvette, France. Her current research interest mainly includes fast algorithms in solving inverse scattering problems, near field imaging, biomedical imaging, and antenna designing.

Séminaire d'Automatique du plateau de Saclay : Computation of Curvature Penalized Shortest Paths via the Fast Marching Algorithm

Séminaire le 28 Février 2017, 11h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Jean-Marie Mirebeau (Laboratoire de mathématique d'Orsay, Université Paris-Sud)


Motivated by applications to motion planning and image segmentation, we consider shortest paths models with a curvature penalization, such as the Euler/Mumford elasticas, or the Reed-Shepp car with or without reverse gear. Our numerical strategy, for computing the path of minimal energy joining two given points, involves approximating these singular models using strongly anisotropic Riemannian or Finslerian metrics on the product space R^d x S^{d-1}. The associated eikonal equations are then solved via specialized variants of the Fast-Marching algorithm.

Bio. Jean-Marie Mirebeau est chargé de recherches au Laboratoire de mathématiques d'Orsay, Université Paris-Sud, CNRS, Université Paris-Saclay. Ses travaux portent sur la résolution numérique des équations aux dérivées partielles, et en particulier sur les difficultés liées aux fortes anisotropies. C'est à dire à l'existence de directions privilégiées par le modèle, non alignées avec les axes de coordonnées. La conception de schémas pour ces modèles requiert des outils mathématiques peu communs en analyse, souvent issus de l'arithmétique et de la géométrie discrète. Son activité englobe l'étude théorique de la convergence et de la complexité des schémas numériques, leur implémentation et leur distribution en license libre, et le suivi de leur application via des collaborations académiques et industrielles. Jean-Marie Mirebeau a reçu le prix Popov 2016 pour ses contributions en théorie de l'approximation. Il était antérieurement affecté au laboratoire Ceremade de l'Université Paris-Dauphine, et a effectué sa thèse sous la direction d'Albert Cohen à l'Université Pierre et Marie Curie.

Séminaire d'Automatique du plateau de Saclay : A relaxation result for state constrained delay differential inclusion

Séminaire le 28 Février 2017, 10h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Ihab Haidar (Laboratoire Quartz, ENSEA-CERGY)


This talk is interested by delay differential inclusions in finite dimensional real space. The celebrated Filippov’s theorem is extended to this case. Then, this theorem is generalized to the case when the state variable is constrained to the closure of an open state subset. Under a new “inward pointing condition”, a relaxation result stating that the set of trajectories lying in the interior of the state constraint is dense in the set of constrained trajectories of the convexified inclusion is shown.
Bio. Ihab Haidar was born in Beirut, Lebanon, in 1983. He received the Master’s degree in mathematics from the University of Aix-Marseille 1, France, in 2008 and the Ph.D degree from the University of Montpellier 2, France, in 2011. Since then he has been post-doc in different places (Laboratoire des Signaux et Systèmes, Institut de Mathématiques de Jussieu, Laboratoire QUARTZ-ENSEA). His research interests include control theory, time delay systems and systems biology.

S³ seminar : FastText: A library for efficient learning of word representations and sentence classification

Séminaire le 24 Février 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Piotr Bojanowski, (Facebook AI Research)


In this talk, I will describe FastText, an open-source library that can be used to train word representations or text classifiers. This library is based on our generalization of the famous word2vec model, allowing to adapt it easily to various applications. I will go over the formulation of the skipgram and cbow models of word2vec and how these were extended to meet the needs of our model. I will describe in details the two applications of our model, namely document classification and building morphologically-rich word representations. In both applications, our model achieves very competitive performance while being very simple and fast.

Séminaire d’Automatique du plateau de Saclay : Cooperative Control of Multi-Agents: On a Sphere Manifold and in the Euclidean Space

Séminaire le 13 Février 2017, 11h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Wei Li (Department of Control and Systems Engineering, Nanjing University)


The talk will discuss cooperative control of multi-agents on a sphere and in the Euclidean space. We will first consider the control law design of agents on a sphere, and analyze the stability, scaling, and geometry properties, and discuss future directions. Then, for agents evolving in the Euclidean space, we will consider coupled agents with second-order dynamics. The state of a single agent includes both position and velocity, thus generally, the agents have both velocity coupling and position couplings (VCPC); and if we consider different VCPC, then interesting yet difficult problems arise. We then discuss two aspects of analysis on consensus convergence , and future directions.

Bio. Wei Li received the Ph.D. degree in Automatic Control from Shanghai Jiao Tong University, Shanghai, China, in 2008.
From 2009 to 2010, he was a Post-Doctoral Research Associate with the Department of Electrical Engineering, The University of Texas at Dallas, Dallas, TX, USA. Since 2010, he has been an Associate Professor with the Department of Control and Systems Engineering, Nanjing University, Nanjing, China. His current research interests include robotics, autonomous mobile robots, decentralized control, cooperative control of mobile robotic agents, and wireless sensor networks. Dr. Li is an Associate Editor of Asian Journal of Control. He is a Senior Member of IEEE.

Séminaire d’Automatique du plateau de Saclay : Observer synthesis under time-varying sampling for Lipschitz nonlinear systems

Séminaire le 13 Février 2017, 10h00 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Lucien Etienne (L2S, CentraleSupelec)


The problem of observation of continuous-time nonlinear Lipschitz systems under time-varying discrete measurements is studied. This class of systems naturally occurs when continuous processes are observed through digital sensors and information is sent via a network to a computer for state estimation. Since network introduces uncertainties in the sampling time, the observer must be designed so to take these uncertainties into account. Here two classes of observation scheme are studied. First an impulsive observers, which make instantaneous correction when information is received, is considered. Then a Luenberger-like observer with a piece wise constant correction term is studied. For both classes of observer, generic conditions are provided. Then a restriction of the generic conditions is used to establish tractable conditions that allows the synthesis of an observer gain.
Bio. Lucien Etienne received a M.Sc. Degree in Applied Mathematics at the INSA Rouen in 2012 and a joint Ph.D. in Automatic Control from the university of L'Aquila and the university of Cergy-Pontoise in 2016. After a Post-doc at INRIA Lille on observer synthesis for sampled data system, he is currently Post-doc at L2S CentralSupéléc working on switched systems for embedded control under mixed stochastic/deterministic timing uncertainty.

His research interests include switched and hybrid systems, Observer synthesis and sampled data systems.

S³ seminar : Stochastic Quasi-Newton Langevin Monte Carlo

Séminaire le 10 Février 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Umut Şimşekli, (LTCI, Télécom ParisTech)


Recently, Stochastic Gradient Markov Chain Monte Carlo (SG-MCMC) methods have been proposed for scaling up Monte Carlo computations to large data problems. Whilst these approaches have proven useful in many applications, vanilla SG-MCMC might suffer from poor mixing rates when random variables exhibit strong couplings under the target densities or big scale differences. In this talk, I will present a novel SG-MCMC method that takes the local geometry into account by using ideas from Quasi-Newton optimization methods. These second order methods directly approximate the inverse Hessian by using a limited history of samples and their gradients. Our method uses dense approximations of the inverse Hessian while keeping the time and memory complexities linear with the dimension of the problem. I will provide formal theoretical analysis where it is shown that the proposed method is asymptotically unbiased and consistent with the posterior expectations. I will finally illustrate the effectiveness of the approach on both synthetic and real datasets. This is a joint work with Roland Badeau, Taylan Cemgil and Gaël Richard. arXiv: https://arxiv.org/abs/1602.03442

S³-PASADENA seminar : Detecting confounding in multivariate linear models via spectral analysis

Séminaire le 31 Janvier 2017, 10h30 à CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Dominik Janzing, Max Planck Institute for Intelligent Systems, Tuebingen, Germany


We study a model where one target variable Y is correlated
with  a vector X:=(X_1,...,X_d) of predictor variables  being potential causes of Y.
We describe  a method that infers to what extent the statistical dependences between X and Y
are due to the influence of X on Y and to what extent due to a hidden common cause
(confounder) of X and Y. The method is based on an independence assumption stating that, in the absence of confounding,
the vector of regression coefficients describing the influence of each X on Y has 'generic orientation'
relative to the eigenspaces  of the covariance matrix of X. For the special case of a scalar confounder we show that confounding typically spoils this generic orientation in a characteristic way that can be used to quantitatively estimate the amount of confounding.
I also show some encouraging experiments with real data, but the method is work in progress and critical comments are highly appreciated.

Postulating 'generic orientation' is inspired by a more general postulate stating that
P(cause) and P(effect|cause) are independent objects of Nature and therefore don't contain information about each other [1,2,3],
an idea that inspired several causal inference methods already, e.g. [4,5].

[1] Janzing, Schoelkopf: Causal inference using the algorithmic Markov condition, IEEE TIT 2010.
[2] Lemeire, Janzing: Replacing causal faithfulness with the algorithmic independence of conditionals, Minds and Machines, 2012.
[3] Schoelkopf et al: On causal and anticausal learning, ICML 2012.
[4] Janzing et al: Telling cause frome effect based on high-dimensional observations, ICML 2010.
[5] Shajarisales et al: Telling cause from effect in deterministic linear dynamical systems, ICML 2015.

Pages