Segmentation-déconvolution d'images texturées: gestion des incertitudes par une approche bayésienne hiérarchique et un échantillonnage stochastique

Seminar on July 09, 2019, 4:30 PM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Jean-François Giovannelli

Résumé : La présentation concerne la déconvolution-segmentation conjointe pour des images présentant des texturées orientées. Les images sont constituées de régions présentant des patchs de textures appartenant à un ensemble de K classes prédéfinies. Chaque classe est modélisée par un champ gaussien piloté par une densité spectrale de puissance paramétrique de paramètres inconnus. Par ailleurs, les labels de classes sont modélisés par un champ de Potts de paramètre est également inconnu. La méthode repose sur une description hiérarchique et une stratégie d'estimation conjointement des labels, des K images texturées, ainsi que des hyperparamètres: niveaux du bruit et des images ainsi que paramètres de texture et du champ de Potts. La stratégie permet de définir des estimateurs optimaux au sens d'un risque joint: maximiseur ou moyenne a posteriori selon les paramètres. Ils sont évalués numériquement à partir d'échantillons de loi a posteriori, eux-mêmes obtenus par un algorithme de Gibbs par bloc. Deux des étapes sont délicates: (1) le tirage des images texturées, gaussiennes de grande dimension, est réalisé par un algorithme de Perturbation-Optimization [a] et (2) le tirage des paramètres des images texturées obtenu par une étape de Fisher Metropolis-Hastings [b]. On donnera plusieurs illustrations numériques notamment en terme de quantification des incertitudes. Le travail est publié dans [c].
[a] F. Orieux, O. Féron and J.-F. Giovannelli, "Sampling high-dimensional Gaussian distributions for general linear inverse problems", Signal Processing Letters, May 2012.
[b] C. Vacar, J.-F. Giovannelli, Y. Berthoumieu, "Langevin and Hessian with Fisher approximation stochastic sampling for parameter estimation of structured covariance" ICASSP 2011.
[b'] M. Girolami, B. Calderhead, "Riemannian manifold Hamiltonian Monte Carlo", Journal of the Royal Statistical Society, 2011.
[c] C. Vacar and J.-F. Giovannelli, "Unsupervised joint deconvolution and segmentation method for textured images: A Bayesian approach and an advanced sampling algorithm", EURASIP Journal on Advances in Signal Processing, 2019

Short bio : Jean-François Giovannelli was born in Beziers, France, in 1966. He received the Dipl. Ing. degree from the Ecole Nationale Supérieure de l'Electronique et de ses Applications, Cergy, France, in 1990, and the Ph.D. degree and the H.D.R. degree in signal-image processing from the Universite Paris-Sud, Orsay, France, in 1995 and 2005, respectively. From 1997 to 2008, he was an Assistant Professor with the Universite Paris-Sud and a Researcher with the Laboratoire des Signaux et Systemes, Groupe Problèmes Inverses. He is currently a Professor with the Universite de Bordeaux, France and a Researcher with the Laboratoire de l'Integration du Matériau au Système, Groupe Signal-Image, France. His research focuses on inverse problems in signal and image processing, mainly unsupervised and myopic problems. From a methodological standpoint, the developed regularization methods are both deterministic (penalty, constraints,...) and Bayesian. Regarding the numerical algorithms, the work relies on optimization and stochastic sampling. His application fields essentially concern astronomical, medical, proteomics, radars and geophysical imaging.

Advances in data processing and machine learning in camera networks

Seminar on July 09, 2019, 3:30 PM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Hichem Snoussi

Résumé : The aim of this tutorial is to give an overview of recent advances in distributed signal/image processing in wireless sensor networks. Over the past few years, wireless sensor networks received tremendous attention for monitoring physical phenomena and for target tracking in a wide region or a critical infrastructure under surveillance. With such systems, the automatic monitoring of an event or an incident is based on the reliability of the network to provide an efficient and robust decision-making. Applying conventional signal/image techniques for distributed information processing is inappropriate for wireless sensor networks, since the computational complexity scales badly with the number of available sensors and their limited energy/memory resources. For this purpose, collaborative information processing in sensor networks is becoming a very attractive field of research. The sensors have the ability to collaborate and exchange information to ensure an optimal decision-making. In this tutorial, we review recently proposed collaborative strategies for self-localization, target tracking and nonlinear functional estimation (nonlinear regression), in a distributed wireless sensor network. The collaborative strategy ensures the efficiency and the robustness of the data processing, while limiting the required communication bandwidth. Signal processing challenges in mobile ad-hoc sensor networks will also be considered in this tutorial.

En compagnie de ses anciens collègues et doctorants, Ali retrace les moments importants de sa carrière.

Seminar on July 09, 2019, 2:00 PM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40

En compagnie de ses anciens collègues et doctorants, Ali retrace les moments importants de sa carrière.

15h/15h30 Pause café devant l'amphi Janet

En compagnie de ses anciens collègues et doctorants, Ali retrace les moments importants de sa carrière.

Seminar on July 09, 2019, 2:00 PM at CentraleSupelec (Gif-sur-Yvette) Amphi Janet

En compagnie de ses anciens collègues et doctorants, Ali retrace les moments importants de sa carrière.

A Bayesian deep learning approach in thermal remote imaging with hyper-resolution

Seminar on July 09, 2019, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Ning Chu

Résumé : Remote monitoring and early warning of thermal source abnormality play more and more important roles in fire prevention for the museums and historical monuments (Notre dame de Paris e.g.), metro and electric vehicle (Tesla e.g.) etc. However, conventional thermal imaging techniques cannot obtain the accurate temperature distribution of thermal sources in the far-fields. This is due to the fact that true temperature of thermal sources, according to heat radiation model, depends on many complex factors such as background temperature, environment humidity and surface emissivity . To solve the above challenge, we propose a Bayesian deep learning approach  in thermal remote imaging with hyper-resolution. And mixture Gaussian priors are employed to model the temperature distribution of thermal sources, as well as background temperature. Meanwhile, sparsity-enforcing prior of temperature gradient is also utilized for spatial hyper-resolution. Moreover, the environment humidity and surface emissivity in heat radiation model can be studied by latent variables in Bayesian Hierarchy Network, so that these two important parameters can be estimated by maximizing the entropy of variational Bayesian inference. Through this Bayesian deep learning framework (sampling-training-updating),  temperature mapping of hot sources can be accurately obtained (about 0.5 degree Celsius variation) as far as 5-10 meters way through a cost-effective infra-red camera (

Short bio : Mr. Ning Chu received the Bachelor in information engineering  from the National University of Defense Technology in 2006. He obtain the master and PhD in automatic signal, and image processing from the University of Paris Sud, France  in 2010 and 2014 respectively. He then won the positions of scientific collaborator in École Polytechnique Fédérale de Lausanne, Switzerland, and senior lecturer in Zhejiang Unviersity. His research interests mainly focus on acoustic source imaging, Bayesian deep learning in condition monitoring and inverse problem applied in super resolution imaging. He has published more than 22 peer-reveiwer journal papers, invited for lectures by top international scientific conferences, own 5 China patents and 6 software copyrights.

The role of blockchain and IoT in the future of smart cities – A practical approach

Seminar on July 04, 2019, 10:00 AM at CentraleSupelec (Gif-sur-Yvette) Salle des séminaires du L2S
Mohammad Hammoudeh

Abstract: The advent of communication technologies and smart storage brought the concept of smart cities. One of the key enabling technologies of smart living is the Internet of Things (IoT). The proliferation in IoT applications raised many serious security concerns to citizens, businesses and governments. Recently, blockchain has been advocated to as a solution for secure data storage and sharing.

In this talk, we explore the challenges of the application of blockchain to IoT using a use case from the pharmaceutical sector. We explore how these leading technologies are making cities smarter by making operations efficient, secure and sustainable. In the second half of the talk, we will deliver a live demonstration of CupCarbon U-One, a smart city simulator.

Bio: Mohammad Hammoudeh​ is a Reader in Future Networks and Security. He is the Head of the CfACS IoT Laboratory within the Department of Computing and Mathematics, Manchester Metropolitan University. He has been a researcher in the field of big sensory data communication, mining and visualisation. He is a highly proficient, experienced, and professionally certified cyber security professional, specialising in threat analysis, and information and network security management. His research interests include highly decentralised algorithms, communication, and cross-layered solutions to Internet of Things, and wireless sensor networks.​

Ahcene Bounceur is an associate professor (HDR and qualified for professorship) of Computer Science and Operations Research at the University of Brest (UBO). He is a member of the Lab-STICC Laboratory. He received a Ph.D. in Micro and Nano electronics at Grenoble INP, France in 2007. He received the M.S. degrees from ENSIMAG, Grenoble, France in 2003. From April 2007 to August 2008, he was a postdoctoral fellow at TIMA Laboratory. From September 2007 to August 2008, he was with Grenoble INP, where he was a temporary professor. He has obtained the 3rd place of the Annual IEEE Test Technology Technical Council (TTTC-IEEE) Doctoral Thesis Contest, Berkeley, May 2007. His current research activities are focused on: Tools for simulation of Wireless Sensor Networks (WSN) dedicated to Smart-cities and IoT, parallel models for accelerating simulations and predicting/testing parameters in WSNs, sampling methods for data mining and Big Data. He is the coordinator of the ANR project PERSEPTEUR and the developer of the IoT platform SUIDIA for gestational diabete monitoring.

Imaging with Electromagnetic Waves and Fields, from Eddy Current to Microwave

Seminar on July 04, 2019, 10:00 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Yu Zhong

Abstract: Imaging problems with electromagnetic waves and fields are of great interest due to non-intrusive inspection enabled by such imaging methods. In this talk, two major imaging methods in two different frequency bands will mainly be discussed, eddy current imaging at low frequency and microwave imaging at resonant frequency regime. As these two types of problems are nonlinear and unstable, from mathematical perspectives, one will show, in each, how these difficulties are specifically handled.

In the first part, the physical mechanism of eddy current inspection will be discussed, followed by a full description of an inspection system. An imaging method that could work with the measured eddy current signals will then be proposed. It includes a forward model for eddy current interactions with defects, an experimental signal calibration model, a defect model for inversion, and an optimization scheme. It will be shown how these bricks work together to provide imaging results from phaseless eddy current signals.
In the second part, the highly nonlinear inverse scattering problems (ISPs) will be shown how to be efficiently tackled by the recently proposed contraction integral equation for inversion (CIE-I), in both three-dimensional (3-D) problems and 2-D problems with phaseless data. With the CIE-I, the non-linearity of ISPs is largely remedied by suppressing multiple scattering effects within the inversions, without compromising the physical model accuracy. This is very important when handling the computationally costly 3-D ISPs, since  each iteration of inversion might cost many computational resources. Compared to conventional imaging methods with the well-known Lippmann-Schwinger integral equation (LSIE), this new imaging method with CIE-I shows much better performance when tackling both 3-D ISPs and 2-D ones with phaseless data, w.r.t. resolvability against non-linearity and convergence speed.


Biography: Yu Zhong received the B.E. and M.E. degrees in electronic engineering from Zhejiang University, Hangzhou, China, in 2003 and 2006, respectively, and the Ph.D. degree in electrical and computer engineering from the National University of Singapore, Singapore, in 2010. He was a Research Engineer and a Fellow with the National University of Singapore, from 2009 to 2013, then involved in a French-Singaporean MERLION Cooperative Program. Since 2014, he has been a Scientist with the Institute of High Performance Computing (IHPC), Agency for Science, Technology and Research, Singapore. He has been regularly invited to the Laboratoire des Signaux et Systèmes (L2S), Gif-sur-Yvette, France, as Senior Scientific Expert once per year since 2012. He was invited as a Visiting Professor to University of Trento, Italy, in June 2018. His current research interests include numerical methods for inverse problems associated with waves and fields, electromagnetic and acoustic modeling with complex materials, and non-destructive testing.

Décrypter le langage sonore des animaux : De la technologie du signal vers l'éthologie animale et l'éthologie homme animal

Seminar on June 14, 2019, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Fabienne DELFOUR (LEEC Paris 13) et Pascal BETREMIEUX (Dolhom)

Résumé: Depuis plusieurs décennies, l'homme a entretenu des relations spéciales avec le monde des cétacés, et certaines espèces en particulier (grand dauphin, cachalot, orques) chez lesquelles il a pu reconnaitre un niveau plus sophistiqué de communication sonore, associé à des comportements sociaux proches de comportements humains. Les principaux signes de ces formes "d'intelligence  supérieure" à d'autres espèces proviennent des stratégies de communication sonore mises en évidence scientifiquement à l'aide de nouvelles technologies de capteurs et d'algorithmes de traitement du signal. Travaillant sur certaines espèces de dauphins, nous exposerons quelques découvertes récentes obtenues à partir d'un système innovant d'observation audio/vidéo 3D, alimentant les débats, plus ouverts que jamais, sur l'intelligence animale.
Nous exposerons également un projet basé sur la définition de nouveaux modes de communications avec les dauphins. Des nouveaux modèles d'interaction homme-animal sont ainsi proposés pour la recherche scientifique, la santé et le grand public. Nous terminerons avec quelques questions autour de la biodiversité et des nouvelles formes de relations qui doivent être inventées entre l'humanité 4.0  et le règne animal.

Biographies: Fabienne  Delfour est chercheuse HDR associée au laboratoire d’éthologie expérimentale et comparée de l’université Paris 13, responsable des programmes scientifiques au delphinarium du parc Astérix et associée au Wild Dolphin Project.
Pascal Bétrémieux, fondateur de la startup Dolhom, s'intéresse aux applications éthologiques et sociétales des récentes découvertes scientifiques autour de l'intelligence des dauphins. Il envisage des applications concrètes dans le domaine de la santé et de la relation homme animal en général.

Probabilité et Mécanique Quantique: Loi de Bayes, Estimation de paramètres

Seminar on May 09, 2019, 11:00 AM at CentraleSupelec (Gif-sur-Yvette) Salle des séminaires du L2S
Clément Pellegrini

Abstract. Dans cet exposé nous reviendrons sur le modèle mathématique décrivant l'expérience de Serge Haroche: "quantum non-demolition experiment" pour lequel il a reçu le prix Nobel de Physique. A travers ce modèle nous verrons comment la loi de Bayes apparait naturellement dans le contexte de la mécanique quantique: notamment dans le contexte des mesures indirectes. Nous verrons ensuite comment nous pouvons faire de l'estimation de paramètres sur ces modèles et comment on peut parler de stabilité du filtre sous-jacent. Cet exposé ne demande pas de prérequis de mécanique quantique, nous introduirons les concepts de base nécessaires.

Bio. Clément Pellegrini, Maitre de conférences à l'université Paul Sabatier Toulouse III depuis 2009
Post-doctorat sous la direction de Francesco Petrucionne à Durban 2008-2009
Doctorat sous la direction de Stéphane Attal à l'université Claude Bernard Lyon: thèse soutenue en 2008

Exponential stabilization of open quantum systems

Seminar on April 26, 2019, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Weichao LIANG

Abstract: In view of the rapid development of quantum information science, the interest in developing systematic theories for controlling quantum systems is rapidly increasing. In particular, the concept of quantum feedback control has central importance in engineering of reliable quantum technologies, similar to the classical case.

In this talk, we will review some basic notations of quantum mechanics and introduce the dynamics of open quantum systems under the continuous-time measurements. We will then present our results on feedback exponential stabilization of Spin-1/2 systems.

Bio: Weichao Liang received the B.Sc degree in Telecommunication from XiDian University, China, in 2014 and the M.Sc. degree in Automatic from CentraleSupélec-Université Paris Sud, France, in 2016. He is currently working toward the Ph.D. degree in Laboratoire des Signaux et Systemes, CentraleSupélec-Université Paris Sud-Université Paris Saclay under the supervision of Paolo Mason and Nina Amini. His research interests include stabilization of open quantum systems, stochastic control and non-linear control.

Control and estimation problems in antilock braking systems

Seminar on April 26, 2019, 10:00 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40

Abstract: Introduced by Bosch in 1978, the antilock braking system (ABS) is nowadays one of the most important safety systems for wheeled vehicles. The ABS has two main objectives: to prevent the wheels from locking during heavy braking in order to maintain the stability and steerability of the vehicle, and to maximally exploit the tyre-road friction coefficient in order to achieve the shortest possible braking distance.

In this talk we will review the basics of the wheel dynamics and discuss three problems related to the ABS. First, we will address the estimation of the so-called extended braking stiffness, which is defined as the derivative of the friction coefficient between the tyre and the road. Next, using this estimation, we will reformulate the (control) objective of the ABS in terms of the extended braking stiffness and present a novel control algorithm for the ABS.

Finally: we will address the estimation of the wheel’s angular velocity and acceleration from the measurements of an incremental encoder with imperfections, a problem which is often overlooked in the literature and whose solution is essential for the operation of the ABS.

Bio: Missie Aguado was born in Mexico City in 1988. She received her B.Sc. degree in electric and electronic engineering in 2012, and her M.Sc. degree in control engineering in 2015, both with the highest honors from the National Autonomous University of Mexico (UNAM). She is currently working towards her PhD degree in automatic control in Univ. Paris-Saclay under the supervision of W. Pasillas-Lépine and Antonio Loría. Her research interests include nonlinear control and estimation with applications to wheeled vehicles and electrical motors.

S³ seminar :Décomposition spectroscopique en imagerie multispectrale

Seminar on April 06, 2018, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Vincent MAZET and Hassan MORTADA (ICube / équipe IMAGeS / groupe IPSEO, Université de Strasbourg)

La cinématique interne des galaxies est une clé pour comprendre l'histoire de l'Univers. Elle peut être étudiée en analysant les raies du spectre de la galaxie qui sont décalées par effet Doppler. Les observations multispectrales des galaxies permettent donc de mesurer le décalage des raies dans chaque pixel. Par ailleurs, la spectroscopie de photoélectrons est une technologie qui permet de suivre l'état d'un système en fonction du temps. Les données produites sont une séquences de spectres dont les raies évoluent au cours des acquisitions. Ces deux applications ont en commun des signaux spectroscopiques, répartis dans l'espace ou le temps, et dont les raies évoluent lentement en longueur d'onde, en intensité et en forme.

Un grand nombre de travaux portent sur la décomposition d'un unique spectre, mais aucune approche ne permet la décomposition simultanée de plusieurs spectres présentant une évolution lente des raies. Le projet DSIM, financé par l'ANR, a permis de développer des outils pour décomposer ces spectres, c'est-à-dire pour estimer le nombre et les paramètres des raies dans les spectres. La décomposition spectroscopique est considérée comme un problème inverse : les raies sont modélisées par une fonction paramétrique dont les paramètres sont à estimer.

Nous avons principalement exploré deux manières d'introduire et de traiter l'information d'évolution lente de ces paramètres. D'une part, le problème a été établi dans le cadre bayésien et l'utilisation de l'algorithme RJMCMC a permis d'obtenir de très bon résultats. D'autre part, afin accélérer le temps de calcul de cette première méthode, nous avons considéré le problème comme une séparation de sources retardées et paramétriques. Le défi réside dans le fait que les sources sont extrêmement corrélées. Un schéma de moindres carrés alternés incluant un algorithme d'approximation parcimonieuse a pour cela été conçu.

Biography: Vincent Mazet a soutenu sa thèse à l'Université de Nancy en 2005. Depuis 2006, il est maître de conférences à l'Université de Strasbourg et effectue ses recherches dans le laboratoire ICube. Ses recherches portent sur les problèmes inverses en traitement d'images, en utilisant en particulier des approches bayésiennes ou par approximation parcimonieuse, et en les appliquant à la spectroscopie, à la télédétection ou à l'imagerie hyperspectrale astronomique.

Hassan Mortada a eu son licence en électronique à l’Université Libanaise (UL) en  2013. Il a obtenu son master en 2015 à l’Université de Brest (master recherche signaux et circuits).  Depuis 2015, il prépare sa thèse à l’Université de Strasbourg, ICUBE. Ses thématiques de recherche concernent les problèmes inverses et l'approximation parcimonieuse appliquée aux données spectroscopiques.

S³ seminar : Chauves-souris, écholocation et neuroscience computationnelle : que nous disent les bornes de Cramer-Rao ?

Seminar on March 09, 2018, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle des séminaires du L2S

L’écholocation chez les mammifères, découverte dans les années 50, n’a pas fini de nous surprendre. L’intérêt pour la discipline, qu’on aborde désormais sur l’angle du sonar/radar cognitif (coté traitement du signal et ingénierie système) ou des neurosciences computationnelles (du côté des biologistes, éthologues ou des neurosciences) semble au contraire connaitre un regain d’intérêt ces dernières années, notamment dans une perspective bayésienne.

Nous montrons dans cet exposé des résultats récents obtenus lors de la mise au point d’un des premiers systèmes opérationnels de géolocalisation acoustique dynamique de l’animal dans son environnement naturel. Dans ce travail, nous exploitons en premier lieu la théorie de Fischer et les célèbres bornes de Cramer-Rao pour affronter, d’une part, la problématique de l’incertitude temps-fréquence intrinsèque aux formes d’onde émises par l’animal et, d’autre part, analyser la problématique de l’adaptation de son système sonar en fonction de l’objectif de perception et des contraintes environnementales.

Ces travaux reprennent les premières tentatives de trajectographie acoustique passive de l’animal par Yves Tupinier et Patrick Flandrin, il y a une quarantaine d’année. Ils dévoilent désormais des résultats concrets particulièrement novateurs sur le plan biologique, comportemental et/ou neurologique. Par ailleurs, la portée industrielle de ces travaux est stratégique à l’heure où nous cherchons désormais à développer des systèmes de drone capable de voler en milieu confiné, ce que la chauve-souris sait faire admirablement, les yeux fermés…

Biography: Didier Mauuary, Ingénieur Centrale Paris (89), spécialité physique de l’océan et de l’atmosphère et Docteur INPG (94) débute ses travaux en acoustique sous-marine pour développer des méthodes d’observation physique globale à l’échelle climatique. Il poursuit ses travaux de recherche en collaboration avec l’université Carnegie Mellon de Pittsburgh et l’institut des sciences de la Mer de Kiel, ce qui l’amène à cosigner un article dans le magazine Nature. Il poursuit ensuite sa carrière dans l’industrie du SONAR, principalement dans le secteur de la Défense et publie une dizaine d’articles scientifiques dans les revues et conférences internationales. Il crée en 2010 la première startup française dont le programme de R&D est principalement axé sur la chauve-souris.

S³ seminar : Fourier transforms of polytopes and their role in Number Theory and Combinatorics

Seminar on December 21, 2017, 11:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Sinai Robins, (University of Sao Paulo, Sao Paulo, Brasil and Brown University, Providence, USA)

We introduce the topic of the Fourier transform of a Euclidean polytope, first by examples and then by more general formulations.  Then we point out how we can use this transform (and the frequency space) to analyze the following problems:
1.  Compute lattice point enumeration formulas for polytopes
2.  Relate the transforms of polytopes to tilings of Euclidean space by translations of a polytope

We will give a flavor of how such applications arise, and we point to some conjectures and applications.

S³ seminar : On the polynomial part of a restricted partition function

Seminar on December 21, 2017, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Karl Dilcher, (Dalhousie University, Halifax, Canada)

We prove an explicit formula for the polynomial part of a restricted
partition function, also known as the first Sylvester wave. This is achieved by way of some identities for higher-order Bernoulli polynomials, one of which is analogous to Raabe's well-known multiplication formula for the ordinary Bernoulli polynomials. As a consequence of our main result we obtain an asymptotic expression of the first Sylvester wave as the coefficients of the restricted partition grow arbitrarily large.
(Joint work with Christophe Vignat).

S³ seminar : Non-negative orthogonal greedy algorithms for sparse approximation

Seminar on December 08, 2017, 10:30 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40

Sparse approximation under non-negativity constraints naturally arises in several applications. Many sparse solvers can be directly extended to the non-negative setting. It is not the case of Orthogonal Matching Pursuit (OMP), a well-known sparse solver, which gradually updates the sparse solution support by selecting a new dictionary atom at each iteration. When dealing with non-negative constraints, the orthogonal projection computed at each OMP iteration is replaced by a non-negative least-squares (NNLS) subproblem whose solution is not explicit. Therefore, the usual recursive (fast) implementations of OMP do not apply. A Non-negative version of OMP (NNOMP) was proposed in the recent literature together with several variations. In my talk, I will first recall the principle of greedy algorithms, in particular NNOMP, and then, I will introduce our proposed improvements, based on the use of the active-set algorithm to address the NNLS subproblems. The structure of the active-set algorithm is indeed intrisically greedy. Moreover, the active-set algorithm can be called with a warm start, allowing us to fastly solve the NNLS subproblems. (Joint work with Charles Soussen (L2S), Jérôme Idier (LS2N), and El-Hadi Djermoune (CRAN).)

Séminaire d'Automatique du Plateau de Saclay : Necessary and sufficient condition for exponential synchronization of nonlinear systems

Seminar on November 30, 2017, 11:00 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Vincent Andrieu (CNRS Researcher, LAGEP-CNRS, Université de Lyon 1, France)

Based on recent works on transverse exponential stability, some necessary and sufficient conditions for the existence of a (locally) exponential synchronizer are established. We show that the existence of a structured synchronizer is equivalent to the existence of a stabilizer for the individual linearized systems (on the synchronization manifold) by a linear state feedback. This, in turns, is also equivalent to the existence of a symmetric covariant tensor field which satisfies a kind of Lyapunov inequality. Based on this property, we provide the construction of such synchronizer. We discuss then the possibility to achieve global synchronization.

Bio. Vincent Andrieu graduated in applied mathematics from “INSA de Rouen”, France, in 2001. After working in ONERA (French aerospace research company), he obtained a PhD degree from “Ecole des Mines de Paris” in 2005. In 2006, he had a research appointment at the Control and Power Group, Dept. EEE, Imperial College London. In 2008, he joined the CNRS-LAAS lab in Toulouse, France, as a “CNRS-chargé de recherche”. Since 2010, he has been working in LAGEP-CNRS, Université de Lyon 1, France. In 2014, he joined the functional analysis group from Bergische Universitäte Wuppertal in Germany, for two sabbatical years. His main research interests are in the feedback stabilization of controlled dynamical nonlinear systems and state estimation problems. He is also interested in practical application of these theoretical problems, and especially in the field of aeronautics and chemical engineering.

Séminaire d'Automatique du Plateau de Saclay : Observer design for nonlinear systems

Seminar on November 30, 2017, 10:00 AM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Pauline Bernard (PhD, PSL Reserch University, Systems and Control Center, MINES ParisTech)

Unlike for linear systems, no systematic method exists for the design of observers for nonlinear systems. However, observer design may be more or less straightforward depending on the coordinates we choose to express the system dynamics. In particular, some specific structures, called normal forms, have been identified for allowing a direct and easier observer construction. It follows that a common way of addressing the problem consists in looking for a reversible change of coordinates transforming the expression of the system dynamics into one of those normal forms, design an observer in those coordinates, and finally deduce an estimate of the system state in the initial coordinates via inversion of the transformation. This talk gives contributions to each of those three steps.
First, we show the interest of a new triangular normal form with continuous (non-Lipschitz) nonlinearities. Indeed, we have noticed that systems which are observable for any input but with an order of differential observability larger than the system dimension, may not be transformable into the standard Lipschitz triangular form, but rather into an "only  continuous" triangular form. In this case, the famous high gain observer no longer is sufficient, and we propose to use  homogeneous observers instead.
Another canonical form of interest is the Hurwitz linear form which admits a trivial observer. The question of transforming a nonlinear system into such a form has only been addressed for autonomous systems with the so-called Lunberger or Kazantzis-Kravaris observers. This design consists in solving a PDE and we show here how it can be extended to time-varying/controlled systems.
As for the inversion of the transformation, this step is far from trivial in practice, in particular when the domain and image spaces have different dimensions. When no explicit expression for a global inverse is available, numerical inversion usually relies on the resolution of a minimization problem with a heavy computational cost. That is why we have developed a method to avoid the explicit inversion of the transformation by bringing the observer dynamics (expressed in the canonical form coordinates) back into the initial system coordinates. This is done by dynamic extension, i.e. by adding some new coordinates to the system and transforming an injective immersion into a surjective diffeomorphism.

Bio. Pauline Bernard graduated from MINES ParisTech in 2014 with a Master degree in Applied Mathematics and Automatic Control. In 2017, she obtained her Ph.D. in Mathematics and Automatic Control at PSL Reserch University, prepared at the Systems and Control Center, MINES ParisTech under the supervision of Laurent Praly and Vincent Andrieu.

Séminaire d’Automatique du plateau de Saclay : Stability analysis of discrete-time infinite-horizon control with discounted cost.

Seminar on November 27, 2017, 3:00 PM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Romain Postoyan (CNRS researcher, Centre de Recherche en Automatique de Nancy)

We analyse the stability of general nonlinear discrete-time systems controlled by an optimal sequence of inputs that minimizes an infinite-horizon discounted cost. First, assumptions related to the controllability of the system and its detectability with respect to the stage cost are made. Uniform semiglobal and practical stability of the closed-loop system is then established, where the adjustable parameter is the discount factor. Stronger stability properties are thereupon guaranteed by gradually strengthening the assumptions. Next, we show that the Lyapunov function used to prove stability is continuous under additional conditions, implying that stability has a certain amount of nominal robustness. The presented approach is flexible and we show that robust stability can still be guaranteed when the sequence of inputs applied to the system is no longer optimal but near-optimal. We also analyse stability for cost functions in which the importance of the stage cost increases with time, opposite to discounting. Finally, we exploit stability to derive new relationships between the optimal value functions of the discounted and undiscounted problems, when the latter is well-defined.

This is a joint work with Lucian Busoniu (TU Cluj, Romania), D. Nesic (University of Melbourne, Australia) and J. Daafouz (CRAN, Université de Lorraine).

Bio. Romain Postoyan received the master degree (``diplôme d'ingénieur'') in Electrical and Control Engineering from ENSEEIHT (France) in 2005. He obtained the M.Sc. by Research in Control Theory & Application from Coventry University (United Kingdom) in 2006 and the Ph.D. in Control Theory from Université Paris-Sud (France) in 2009. In 2010, he was a research assistant at the University of Melbourne (Australia). Since 2011, he is a CNRS researcher at the Centre de Recherche en Automatique de Nancy (France). He serves as an Associate Editor at the Conference Editorial Board of the IEEE Control Systems Society and for the journals: Automatica, IEEE Control Systems Letters, and IMA Journal of Mathematical Control and Information.

S³ seminar : A Random Block-Coordinate Douglas-Rachford Splitting Method with Low Computational Complexity for Binary Logistic Regression

Seminar on November 24, 2017, 2:00 PM at CentraleSupelec (Gif-sur-Yvette) Salle du conseil du L2S - B4.40
Émilie Chouzenoux (CVN, CentraleSupélec/INRIA, Université Paris-Est Marne-La-Vallée)

In this talk, I will present a new optimization algorithm for sparse logistic regression based on a stochastic version of the Douglas-Rachford splitting method. The algorithm sweeps the training set by randomly selecting a mini-batch of data at each iteration, and it allows us to update the variables in a block coordinate manner. Our approach leverages the proximity operator of the logistic loss, which is expressed with the generalized Lambert W function. Experiments carried out on standard datasets demonstrate the efficiency of our approach w.r.t. stochastic gradient-like methods. (joint work with Luis M. Briceño-Arias, Afef Cherni, Giovanni Chierchia and Jean-Christophe Pesquet)