Abstract
We consider the problem of sampling from a target distribution which is\emph{not necessarily logconcave}. Non-asymptotic analysis results areestablished in a suitable Wasserstein-type distance of the Stochastic GradientLangevin Dynamics (SGLD) algorithm, when the gradient is driven by even\emph{dependent} data streams. Our estimates are sharper and \emph{uniform} inthe number of iterations, in contrast to those in previous studies.
Quick Read (beta)
loading the full paper ...