13. Prosinec 2019
Iterative Bregman Projections algorithm (IBP) proposed by Benamou et. al. in 2015 is a generalization of the well-known Sinkhorn algorithm. It is used to solve a wide range of computational problems related to the optimal transport with the entropic regularisation. In this talk we discuss the convergence rates of IBP applied to computation of Wasserstein barycenters, and classical Monge-Kantorovich problem.
Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes invited talks, demonstrations, symposia and oral and poster presentations of refereed papers. Following the conference, there are workshops which provide a less formal setting.
Pro uložení prezentace do věčného trezoru hlasovalo 1 diváků, což je 0.1 %
Prezentace na podobné téma, kategorii nebo přednášejícího