Fast Convergence for Langevin Diffusion with Manifold Structure

Publication date: 13 February 2020

Abstract: In this paper, we study the problem of sampling from distributions of the form p(x) propto e^{-�eta f(x)} for some function f whose values and gradients we can query. This mode of access to f is natural in the scenarios in which such problems arise, for instance sampling from posteriors in parametric Bayesian models. Classical results show that a natural random walk, Langevin diffusion, mixes rapidly when f is convex. Unfortunately, even in simple examples, the applications listed above will entail working with functions f that are nonconvex -- for which sampling from p may in general require an exponential number of queries. In this paper, we focus on an aspect of nonconvexity relevant for modern machine learning applications: existence of invariances (symmetries) in the function f, as a result of which the distribution p will have manifolds of points with equal probability. First, we give a recipe for proving mixing time bounds for Langevin diffusion as a function of the geometry of these manifolds. Second, we specialize our arguments to classic matrix factorization-like Bayesian inference problems where we get noisy measurements A(XX^T), X in R^{d imes k} of a low-rank matrix, i.e. f(X) = |A(XX^T) - b|^2_2, X in R^{d imes k}, and �eta the inverse of the variance of the noise. Such functions f are invariant under orthogonal transformations, and include problems like matrix factorization, sensing, completion. Beyond sampling, Langevin dynamics is a popular toy model for studying stochastic gradient descent. Along these lines, we believe that our work is an important first step towards understanding how SGD behaves when there is a high degree of symmetry in the space of parameters the produce the same output.

This page was built for publication: Fast Convergence for Langevin Diffusion with Manifold Structure

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6334752)