ICML 2012 Workshop: Sparsity, Dictionaries and Projections in Machine Learning and Signal Processing

Remi Gribonval let us know of the following: potential very interesting workshop where sparsity, compressed sensing, signal processing and matrix factorization are bound to collide:

====================================================================

                         CALL FOR PAPERS

Sparsity, Dictionaries and Projections in Machine Learning and Signal Processing

             ICML 2012 Workshop, Edinburgh, Scotland

                    30 June or 1 July, 2012

     http://www.di.ens.fr/~obozinski/ICML2012workshop/

        Submission Deadline: Monday, May 7, 2012 (5:00 pm PDT)

====================================================================

Overview

Sparse representations are today key in many fields of applied mathematics faced with data, from signal processing to machine learning and statistics.

Historically, several communities proposed various approaches to sparse coding: on the one hand the use of carefully crafted dictionaries, like wavelets, forming bases of functional spaces with good approximation properties over a class of signals, but constructed without data; on the other hand, the use of representations derived directly from data via either algebraic formulations like sparse matrix factorization, or via probabilistic formulations, based on the introduction of latent variables, such as, e.g., independent component analysis or latent Dirichlet allocation.

The introduction of sparsity and/or structure in matrix factorization scheme, which where previously used for dimensionality reduction, induced major shifts in several existing paradigms and led to significant breakthroughs, which have demonstrated the ability of sparse models to provide concise descriptions of certain high dimensional data through low-dimensional projections, together with algorithms of provable performance and bounded complexity. Compressed sensing (and more generally the clever use of random low-dimensional projections), dictionary learning, and non-parametric topic models, are just a few of the rapidly emerging paradigms in this area at the

confluence of signal processing and machine learning.

While sparse models and random low-dimensional projections are already at the heart of several success stories in signal processing and machine learning, their full potential is yet to be achieved and calls for further understanding. The goal of the workshop is to confront the various point of views and foster exchanges of ideas between the signal processing, statistics, machine learning and applied mathematics communities.

We encourage submissions exploring various aspects of learning sparse models and/or latent representations, in the form of new algorithms, theoretical advances and/or empirical results. Some specific areas of interest include structured matrix factorization algorithms, Bayesian models for latent variable representations, analysis of random dictionaries versus learned dictionaries, novel applications of dictionary learning or relationships to compressed sensing.

Submission Guidelines

Submissions should be written as extended abstracts, no longer than 4 pages in the ICML latex style. Style files and formatting instructions can be found at http://icml.cc/2012//files/icml2012stylefiles.zip.

Submissions must be in PDF format. Authors' names and affiliations should be included, as the review process will not be double blind. The extended abstract may be accompanied by an unlimited appendix and other supplementary material, with the understanding that anything

beyond 4 pages may be ignored by reviewers.

Please send your PDF submission by email to guillaume.obozinski@ens.fr with the words "ICML workshop submission" in the title by 5:00 pm PDT on Monday, May 7. Notifications will be given on or before 21 May. Please include these words "ICML workshop" in the title of mails you

are sending about the workshop or workshop submissions. Work that is pending review, was recently published or was presented elsewhere will be considered, provided that the extended abstract mentions this explicitly. Finally, note that there will be no official proceedings

from this workshop.

Invited Speakers (to be confirmed)

Pierre Comon (CNRS / University Nice Sophia Antipolis)

Julien Mairal (University of California at Berkeley)

Matthias Seeger (Ecole Polytechnique Federale de Lausanne)

Daniel Vainsencher (Technion - Israel Institute of Technology)

Organizers

Michael Davies (Edinburgh), Rémi Gribonval (INRIA), Rodolphe Jenatton

(CNRS) and Guillaume Obozinski (INRIA)

Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.