Skip to yearly menu bar Skip to main content

Spotlight Poster

Provably Fast Finite Particle Variants of SVGD via Virtual Particle Stochastic Approximation

Aniket Das · Dheeraj Nagaraj

Great Hall & Hall B1+B2 (level 1) #1206
[ ]
Wed 13 Dec 8:45 a.m. PST — 10:45 a.m. PST

Abstract: Stein Variational Gradient Descent (SVGD) is a popular particle-based variational inference algorithm with impressive empirical performance across various domains. Although the population (i.e, infinite-particle) limit dynamics of SVGD is well characterized, its behavior in the finite-particle regime is far less understood. To this end, our work introduces the notion of *virtual particles* to develop novel stochastic approximations of population-limit SVGD dynamics in the space of probability measures, that are exactly realizable using finite particles. As a result, we design two computationally efficient variants of SVGD, namely VP-SVGD and GB-SVGD, with provably fast finite-particle convergence rates. Our algorithms can be viewed as specific random-batch approximations of SVGD, which are computationally more efficient than ordinary SVGD. We show that the $n$ particles output by VP-SVGD and GB-SVGD, run for $T$ steps with batch-size $K$, are at-least as good as i.i.d samples from a distribution whose Kernel Stein Discrepancy to the target is at most $O(\tfrac{d^{1/3}}{(KT)^{1/6}})$ under standard assumptions. Our results also hold under a mild growth condition on the potential function, which is much weaker than the isoperimetric (e.g. Poincare Inequality) or information-transport conditions (e.g. Talagrand's Inequality $\mathsf{T}_1$) generally considered in prior works. As a corollary, we analyze the convergence of the empirical measure (of the particles output by VP-SVGD and GB-SVGD) to the target distribution and demonstrate a **double exponential improvement** over the best known finite-particle analysis of SVGD. Beyond this, our results present the **first known oracle complexities for this setting with polynomial dimension dependence**, thereby completely eliminating the curse of dimensionality exhibited by previously known finite-particle rates.

Chat is not available.