You are here

Empirical estimation of Wasserstein-1 distance on a Riemannian manifold under group symmetry.

Event Category:
Learning Learning
Ziyu Chen

I will introduce a recent work in sample complexity analysis for probabilistic divergences, in particular, Wasserstein-1 metric, when the distribution is G-invariant and lies on a compact, connected and boundary-less Riemannian manifold. The theory is based on quotient spaces and eigenvalues/eigenfunctions of the Laplace-Beltrami operator.

Thursday, April 4, 2024 - 11:30am
LGRT 1621