arXiv Analytics

Sign in

arXiv:1803.07033 [math.OC]AbstractReferencesReviewsResources

Natural gradient via optimal transport I

Wuchen Li, Guido Montufar

Published 2018-03-16Version 1

We study a natural Wasserstein gradient flow on manifolds of probability distributions with discrete sample spaces. We derive the Riemannian structure for the probability simplex from the dynamical formulation of the Wasserstein distance on a weighted graph. We pull back the geometric structure to the parameter space of any given probability model, which allows us to define a natural gradient flow there. In contrast to the natural Fisher-Rao gradient, the natural Wasserstein gradient incorporates a ground metric on sample space. We discuss implementations following the forward and backward Euler methods. We illustrate the analysis on elementary exponential family examples.

Related articles: Most relevant | Search more
arXiv:1808.05064 [math.OC] (Published 2018-08-15)
On optimal transport of matrix-valued measures
arXiv:1909.11716 [math.OC] (Published 2019-09-25)
Optimal Transport to a Variety
arXiv:1910.09618 [math.OC] (Published 2019-10-21)
Geometry of Graph Partitions via Optimal Transport