Source author record

Flavien Léger

Explore this author’s publications, collaborators and research topics. This catalog record has not been claimed by its author.

ResearcherUnclaimed source record

math.OC Machine Learning math-ph math.AP math.MP math.NA Numerical Analysis

Catalog footprint

Research overview

4works

7topics

3close collaborators

Actions

Is this your work?

Open graph Browse papers

preprint2020arXiv

A fast approach to optimal transport: The back-and-forth method

We present an iterative method to efficiently solve the optimal transportation problem for a class of strictly convex costs which includes quadratic and p-power costs. Given two probability measures supported on a discrete grid with n points, we compute the optimal map using O(n) storage space and O(n log(n)) operations per iteration, with an approximately exponential convergence rate. Our approach allows us to solve optimal transportation problems on spatial grids as large as 4096x4096 and 384x384x384 in a matter of minutes.

preprint2020arXiv

A gradient descent perspective on Sinkhorn

We present a new perspective on the popular Sinkhorn algorithm, showing that it can be seen as a Bregman gradient descent (mirror descent) of a relative entropy (Kullback-Leibler divergence). This viewpoint implies a new sublinear convergence rate with a robust constant.

preprint2016arXiv

A new approach to bounds on mixing

We consider mixing by incompressible flows. In 2003, Bressan stated a conjecture concerning a bound on the mixing achieved by the flow in terms of an $L^1$ norm of the velocity field. Existing results in the literature use an $L^p$ norm with $p>1$ . In this paper we introduce a new approach to prove such results. It recovers most of the existing results and offers new perpective on the problem. Our approach makes use of a recent harmonic analysis estimate from Seeger, Smart and Street.

preprint2010arXiv

Efficient Matrix Completion with Gaussian Models

A general framework based on Gaussian models and a MAP-EM algorithm is introduced in this paper for solving matrix/table completion problems. The numerical experiments with the standard and challenging movie ratings data show that the proposed approach, based on probably one of the simplest probabilistic models, leads to the results in the same ballpark as the state-of-the-art, at a lower computational cost.

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.