Previous | 
Home
Section: Partnerships and Cooperations
National Initiatives
DGA/Thales
Participants : Emmanuel Duflos, Philippe Vanheeghe, Emmanuel Delande.
Title: Multi-sensor PHD filtering with application to sensor management (http://www.theses.fr/2012ECLI0001 )
Web site: http://www.theses.fr/2012ECLI0001
Abstract: The defense of this PhD thesis was held in January 2012.
ANR-Lampada
Participants : Mohammad Ghavamzadeh, Hachem Kadri, Jérémie Mary, Olivier Nicol, Philippe Preux, Daniil Ryabko, Christophe Salperwyck.
Title: Learning Algorithms, Models an sPArse representations for structured DAta
Others partners: Laboratoire d'Informatique Fondamentale de Marseille, Laboratoire Hubert Curien ; Saint Etienne, Laboratoire d'Informatique de Paris 6.
Web site: http://lampada.gforge.inria.fr/
Abstract: Lampada is a fundamental research project on machine learning and structured data. It focuses on scaling learning algorithms to handle large sets of complex data. The main challenges are 1) high dimension learning problems, 2) large sets of data and 3) dynamics of data. Complex data we consider are evolving and composed of parts in some relations. Representations of these data embed both structure and content information and are typically large sequences, trees and graphs. The main application domains are web2, social networks and biological data.
The project proposes to study formal representations of such data together with incremental or sequential machine learning methods and similarity learning methods.
The representation research topic includes condensed data representation, sampling, prototype selection and representation of streams of data. Machine learning methods include edit distance learning, reinforcement learning and incremental methods, density estimation of structured data and learning on streams.
Activity Report: Philippe Preux has collaborated with Ludovic Denoyer and Gabriel Dulac-Arnold from LIP'6 to investigate further the idea of datum-wise representation, introduced in 2011, and originally published at ECML/PKDD'2011. This eventually led to a deeped presentation in the Machine Learning Journal.
They also studied the reinforcement learning problem in the case of a large but not infinite number of actions (hundreds, or thousands discrete actions). They introduced the use of Error-correcting output codes to deal with this setting, proposed, and studied two RL algorithms that take advantage of an ECOC-based representation of actions. The idea was published at ECML/PKDD'2012 and other conferences (EWRL workshop held as part of the ICML conference, and French ones).
Hachem Kadri and Philippe Preux have continued their work on machine learning for functional data. They introduced an algorithm for multiple operators learning. Along with Mohammad Ghavamzadeh, they only introduced a operator-based aprroach for structured output.
Danil Ryabko and colleagues have obtained new results on nonparametric clustering of time-series data. In particular, a fully online clustering algorithm has been developed; we have also shown how to use binary classification methods for clustering time series.
ANR EXPLO-RA
Participants : Alexandra Carpentier, Mohammad Ghavamzadeh, Jean-François Hren, Alessandro Lazaric, Rémi Munos, Daniil Ryabko.
Title: EXPLOration - EXPLOitation for efficient Resource Allocation with Applications to optimization, control, learning, and games
Others partners: Inria Saclay - Ile de France (TAO), HEC Paris (GREGHEC), Ecole Nationale des Ponts et Chaussées (CERTIS), Université Paris 5 (CRIP5), Université Paris Dauphine (LAMSADE).
Activity Report: We developed bandit algorithm for planning in Markov Decision Processes based on the optimism in the face of uncertainty principle.
ANR CO-ADAPT
Participants : Alexandra Carpentier, Rémi Munos.
Other Partners: Inria Odyssee project (Maureen Clerc), the INSERM U821 team (Olivier Bertrand), the Laboratory of Neurobiology of Cognition (CNRS) (Boris Burle) and the laboratory of Analysis, topology and probabilities (CNRS and University of Provence) (Bruno Torresani).
Web site: https://twiki-sop.inria.fr/twiki/bin/view/Projets/Athena/CoAdapt/WebHome
Abstract: The aim of CoAdapt is to propose new directions for BCI design, by modeling explicitly the co-adaptation taking place between the user and the system. The goal of CoAdapt is to study the co-adaptation between a user and a BCI system in the course of training and operation. The quality of the interface will be judged according to several criteria (reliability, learning curve, error correction, bit rate). BCI will be considered under a joint perspective: the user's and the system's. From the user's brain activity, features must be extracted, and translated into commands to drive the BCI system. From the point of view of the system, it is important to devise adaptive learning strategies, because the brain activity is not stable in time. How to adapt the features in the course of BCI operation is a difficult and important topic of research. We will investigate Reinforcement Learning (RL) techniques to address the above questions.
Activity Report: See https://twiki-sop.inria.fr/twiki/bin/view/Projets/Athena/CoAdapt/WebHome
ANR AMATIS
Participant : Pierre Chainais.
Title: Multifractal Analysis and Applications to Signal and Image Processing
Other Partners: Univ. Paris-Est Créteil, Univ. Sciences et Technologies de Lille and Inria (Lille, ENST (Telechom ParisTech), Univ. Blaise Pascal (Clermont-Ferrand), and Univ. Bretagne Sud (Vannes), Statistical Signal Processing group at the Physics Department at the Ecole Normale Supérieure de Lyon, one researcher from the Math. Department of Institut National des Sciences Appliquees de Lyon and two researchers from the Laboratoire d'Analyse, Topologie et Probabilités (LAPT) of Aix-Marseille University.
Abstract: Multifractal analysis refers to two concepts of different natures : On the theoretical side, it corresponds to pointwise singularity characterization and fractional dimension determination ; on the applied side, it is associated with scale invariance characterization, involving a family of parame- ters, the scaling function, used in classification or model selection. Following the seminal ideas of Parisi and Frisch in the mid-80s, these two components are usually related by a Legendre transform, stemming from a heuristic argument relying on large deviation and statistical thermodynamics prin- ciples : The multifractal formalism. This led to new theoretical approaches for the study of singula- rities of functions and measures, as well as efficient tools for classification and models selection, that allowed to settle longstanding issues (e.g., concerning the modeling of fully developed turbulence). Though this formalism had been shown to hold for large classes of functions of widely different origins, the generality of its level of validity remains an open issue. Despite its popularity in appli- cations, the interactions between theoretical developments and applications are unsatisfactory. Its use in image processing for instance is still in its infancy. This is partly due to discrepancy between the theoretical contributions mostly grounded in functional analysis and geometric measure theory, and applications naturally implying a stochastic or statistical framework. The AMATIS project aims at addressing these issues, by proposing a consistent and documented framework combining different theoretical approaches and bridging the gap towards applications. To that end, it will both address a number of challenging theoretical issues and devote significant efforts to elaborating a WEB platform with softwares and documentation. It will combine the efforts of mathematicians with those of physicists and experts in signal and image processing. Dissemination among and interactions between scientific fields are also intended via the organization of summer schools and workshop.
Activity Report: a collaboration with P. Bas (CR CNRS, LAGIS) has started on the steganalysis of textured images. While steganography aims at hiding a message within some support, e.g. a numerical image, steganalysis aims at detecting the presence or not of any hidden message in the support. Steganalysis involves two main tasks: first identify relevant features which may be sensitive to the presence of a hidden message, then use supervised classification to build a detector. While the steganalysis of usual images has been well studied, the case of textured images, for which multifractal models may be relevant, is much more difficult. Indeed, textured images have a rich and disordered content which favors hiding information in an unperceptible manner. A student internship of 6 months at Master level has finished in November. The purpose was to explore the potential of new multiscale wavelet based discriminant features for steganalysis.
National Partners
Inria Nancy - Grand Est, Team MAIA, France.
-
We have had collaboration on the topics of approximate dynamic programming and statistical learning and high-dimensional reinforcement learning this year. On the first topic, we have published a conference paper [47] and a technical report [62] , and on the second one we have published a conference paper [36] together.
-
Supélec, IMS Research Group, Metz, France.
-
We have had collaboration on the topics of approximate dynamic programming and statistical learning and high-dimensional reinforcement learning this year. On the first topic, we have published a conference paper [47] and a technical report [62] , and on the second one we have published a conference paper [36] together.
-
-
-
We have a collaboration on the topic of reinforcement learning, sparse representation. We have worked on the datum-wise representation of data, as well as the handling of large but non infinite sets of actions. See section 8.2.2 for further details.
-