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

• URL: http://arxiv.org/abs/2305.17558v4
• Date: Thu, 5 Oct 2023 18:59:38 GMT
• Title: Provably Fast Finite Particle Variants of SVGD via Virtual Particle Stochastic Approximation
• Authors: Aniket Das and Dheeraj Nagaraj
• Abstract summary: Stein Variational Gradient Descent (SVGD) is a popular variational inference which simulates an interacting particle system to approximately sample from a target distribution. We introduce the notion of virtual particles and develop novel approximations of population-limit dynamics in the space of probability measures. We show that the $n$ particles output by VP-SVGD and GB-SVGD, run for $T$ steps with batch-size $K$, are as good as i.i.i.d samples from a distribution whose Kernel Stein Discrepancy to the target is at most $Oleft(tfrac
• Score: 9.065034043031668
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Stein Variational Gradient Descent (SVGD) is a popular variational inference algorithm which simulates an interacting particle system to approximately sample from a target distribution, with impressive empirical performance across various domains. Theoretically, its population (i.e, infinite-particle) limit dynamics is well studied but the behavior of SVGD in the finite-particle regime is much less understood. In this work, we design two computationally efficient variants of SVGD, namely VP-SVGD and GB-SVGD, with provably fast finite-particle convergence rates. We introduce the notion of virtual particles and develop novel stochastic approximations of population-limit SVGD dynamics in the space of probability measures, which are exactly implementable using a finite number of particles. 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 $O\left(\tfrac{d^{1/3}}{(KT)^{1/6}}\right)$ 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 consider 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.

Related papers

• Long-time asymptotics of noisy SVGD outside the population limit [9.2081159465248]
  We study the long-time behavior of a noisy variant of Stein Variational Gradient Descent (SVGD) In particular, noisy SVGD provably avoids the variance collapse observed for SVGD. Our approach involves demonstrating that the trajectories of noisy SVGD closely resemble those described by a McKean-Vlasov process.
  arXiv  Detail & Related papers  (2024-06-17T13:00:51Z)
• Analytic-Splatting: Anti-Aliased 3D Gaussian Splatting via Analytic Integration [49.004898985671815]
  3DGS is not alias-free, and its rendering at varying resolutions could produce severe blurring or jaggies. This is because 3DGS treats each pixel as an isolated, single point rather than as an area, causing insensitivity to changes in the footprints of pixels. We introduce this approximation in the two-dimensional pixel shading, and present Analytic-Splatting, which analytically approximates the Gaussian integral within the 2D-pixel window area.
  arXiv  Detail & Related papers  (2024-03-17T02:06:03Z)
• Closed-form Filtering for Non-linear Systems [83.91296397912218]
  We propose a new class of filters based on Gaussian PSD Models, which offer several advantages in terms of density approximation and computational efficiency. We show that filtering can be efficiently performed in closed form when transitions and observations are Gaussian PSD Models. Our proposed estimator enjoys strong theoretical guarantees, with estimation error that depends on the quality of the approximation and is adaptive to the regularity of the transition probabilities.
  arXiv  Detail & Related papers  (2024-02-15T08:51:49Z)
• Towards Understanding the Dynamics of Gaussian-Stein Variational Gradient Descent [16.16051064618816]
  Stein Variational Gradient Descent (SVGD) is a nonparametric particle-based deterministic sampling algorithm. We study the dynamics of the Gaussian-SVGD projected to the family of Gaussian distributions via the bilinear kernel. We propose a density-based and a particle-based implementation of the Gaussian-SVGD, and show that several recent algorithms for GVI, proposed from different perspectives, emerge as special cases of our unified framework.
  arXiv  Detail & Related papers  (2023-05-23T13:55:47Z)
• Augmented Message Passing Stein Variational Gradient Descent [3.5788754401889014]
  We study the isotropy property of finite particles during the convergence process. All particles tend to cluster around the particle center within a certain range. Our algorithm achieves satisfactory accuracy and overcomes the variance collapse problem in various benchmark problems.
  arXiv  Detail & Related papers  (2023-05-18T01:13:04Z)
• A Finite-Particle Convergence Rate for Stein Variational Gradient Descent [47.6818454221125]
  We provide the first finite-particle convergence rate for Stein variational descent gradient (SVGD) Our explicit, non-asymptotic proof strategy will serve as a template for future refinements.
  arXiv  Detail & Related papers  (2022-11-17T17:50:39Z)
• Kernel Stein Generative Modeling [68.03537693810972]
  Gradient Langevin Dynamics (SGLD) demonstrates impressive results with energy-based models on high-dimensional and complex data distributions. Stein Variational Gradient Descent (SVGD) is a deterministic sampling algorithm that iteratively transports a set of particles to approximate a given distribution. We propose noise conditional kernel SVGD (NCK-SVGD), that works in tandem with the recently introduced Noise Conditional Score Network estimator.
  arXiv  Detail & Related papers  (2020-07-06T21:26:04Z)
• A Non-Asymptotic Analysis for Stein Variational Gradient Descent [44.30569261307296]
  We provide a novel finite time analysis for the Stein Variational Gradient Descent algorithm. We provide a descent lemma establishing that the algorithm decreases the objective at each iteration. We also provide a convergence result of the finite particle system corresponding to the practical implementation of SVGD to its population version.
  arXiv  Detail & Related papers  (2020-06-17T12:01:33Z)
• Stein Variational Inference for Discrete Distributions [70.19352762933259]
  We propose a simple yet general framework that transforms discrete distributions to equivalent piecewise continuous distributions. Our method outperforms traditional algorithms such as Gibbs sampling and discontinuous Hamiltonian Monte Carlo. We demonstrate that our method provides a promising tool for learning ensembles of binarized neural network (BNN) In addition, such transform can be straightforwardly employed in gradient-free kernelized Stein discrepancy to perform goodness-of-fit (GOF) test on discrete distributions.
  arXiv  Detail & Related papers  (2020-03-01T22:45:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.