Workshop
"Greedy methods meet multiview learning"
22th of september 2014
Overview
This workshop aims at enabling two communities - people working on greedy methods and those interested in multiview machine learning - to meet, exchange ideas and possibly start collaborating.
Greedy methods are powerful tools, making it possible to solve complex problems with simple algorithms, that are provably reliable. Although not recent, this field of study is quite active, in particular in link with recent the recent field of sparse optimization (e.g. with Compressed Sensing).
Data are said to be multiview when several descriptors of different nature are available (e.g. describe a dog from sound and image). Multiview learning is a recent domain of machine learning that aims at exploiting this diversity of descriptors - called views - in particular when there exists no mechanism to transfer the information from one view to another.
Both domains tackle common current issues in signal and image processing, such as the need to handle huge dimensions. Greedy methods have proved to be effective for dealing with high dimensional data, while multiview learning offers a new paradigm for managing diversity. This workshop aims at presenting recent results, identifying common issues and possibly create new collaborations.
Venue
Program
Confirmed speakers
Xavier Bresson, EPFL, Switzerland.
Victor Chepoi, LIF, Marseille.
Laurent Daudet, Institut Langevin, Paris.
François-Xavier Dupé, LIF, Marseille.
Isabelle Guyon, Clopinet, USA.
Ludmila Kuncheva, Bangor University, UK.