NIPS 2014 Workshop - Optimal Transport & Machine Learning
1 Septembre 2014
Catégorie : Conférence internationale
First Call for Contributions -- NIPS 2014 Workshop
OTML 2014 - Optimal Transport & Machine Learning
December 12 or 13, 2014
Montréal, Quebec Canada
Optimal transport (OT) has emerged as a novel tool to solve problems in machine learning and related fields, e.g. graphics, statistics, data analysis, computer vision, economics and imaging.
In particular, the toolbox of OT (including for instance theWasserstein/Earth Mover's Distances) offers robust mathematical techniques to study probability measures and compare complex objects described using bags-of-features representations.
Scaling OT algorithms to datasets of large dimension and sample size presents, however, a considerable computational challenge. Taking for granted that these challenges are partially solved, there remains many salient open research questions on how to integrate OT in statistical methodologies (dimensionality reduction, inference, modeling) beyond its classical use in retrieval. OTML 2014 will be the first international workshop to address state-of-the-art research in this exciting area.
Topics Covered by the Workshop
We solicit submission of original research at the interface between optimal transport and machine learning, including (but not limited to)
- numerical schemes to solve the OT problem (e.g. fast EMD solvers);
- generalizations of the OT problem (e.g. multi-marginal OT);
- theoretical results on approximations of OT distances through embeddings;
- numerical resolution and theoretical study of constrained and/or regularized OT problems;
- applications of OT in supervised learning settings to histogram data (e.g. distance or kernel based classifiers, metric learning with the OT geometry);
- applications of OT in unsupervised learning to summarizing datasets (e.g. generalization of k-means type clustering) and/or histograms (e.g. Wasserstein barycenters, Wasserstein propagation, retrieval);
- applications of OT to modeling interactions in social networks (e.g. matching models);
- applications of OT to computer vision and related fields; and
- combinatorial perspectives on OT and related algebraic statistical concepts (e.g. contingency tables).
- Alexander Barvinok, University of Michigan
- Piotr Indyk, Massachusetts Institute of Technology
- Robert McCann, University of Toronto (to be confirmed)
Submission deadline: October 23rd, 2014
Acceptance decision: October 31st, 2014
Workshop: December 12th or 13th, 2014
Submissions must adhere to NIPS 2014 style format available on the NIPS submission page. Papers may be between 6 and 10 pages long, including figures and references. Supplementary material can be provided. Submissions should not be anonymized.
Submissions should be uploaded on the OTML14 Easychair system before October 23, 23:59PM (PDT, Pacific Time).
All papers will undergo review. All accepted papers will have a poster presentation at the workshop, and selected papers will also be presented orally.
Marco Cuturi, Kyoto University
Gabriel Peyré, CNRS & CEREMADE, Paris Dauphine
Justin Solomon, Stanford University