Michael I. Jordan

TL;DR: This paper showed that the population likelihood function has bad local maxima even in the special case of equally-weighted mixtures of well-separated and spherical Gaussians, showing that the log-likelihood value of these bad local minima can be arbitrarily worse than that of any global optimum.

TL;DR: In this paper, a new approach for comparing reinforcement learning policies, using Wasserstein distances (WDs) in a newly defined latent behavioral space, is introduced. And the dual formulation of the WD is used to learn score functions over policy behaviors that can in turn be used to lead policy optimization towards (or away from) (un)desired behaviors.

TL;DR: In this article, a combination of graph partitioning algorithms with a generalized mean field (GMF) inference algorithm is presented, which optimizes over disjoint clustering of variables and performs inference using those clusters.

TL;DR: Divide-Factor-Combine (DFC) as mentioned in this paper divides a large-scale matrix factorization task into smaller subproblems, solves each subproblem in parallel using an arbitrary base matrix factorisation algorithm, and combines the sub-problem solutions using techniques from randomized matrix approximation.

TL;DR: In this article, a semi-supervised variant of scVI, called Single-cell ANnotation using Variational Inference (scANVI), was proposed to leverage any available cell state annotations.