Score-Based Diffusion meets Annealed Importance Sampling

08/16/2022
by   Arnaud Doucet, et al.
0

More than twenty years after its introduction, Annealed Importance Sampling (AIS) remains one of the most effective methods for marginal likelihood estimation. It relies on a sequence of distributions interpolating between a tractable initial distribution and the target distribution of interest which we simulate from approximately using a non-homogeneous Markov chain. To obtain an importance sampling estimate of the marginal likelihood, AIS introduces an extended target distribution to reweight the Markov chain proposal. While much effort has been devoted to improving the proposal distribution used by AIS, by changing the intermediate distributions and corresponding Markov kernels, an underappreciated issue is that AIS uses a convenient but suboptimal extended target distribution. This can hinder its performance. We here leverage recent progress in score-based generative modeling (SGM) to approximate the optimal extended target distribution for AIS proposals corresponding to the discretization of Langevin and Hamiltonian dynamics. We demonstrate these novel, differentiable, AIS procedures on a number of synthetic benchmark distributions and variational auto-encoders.

READ FULL TEXT
research
07/17/2022

The Importance Markov Chain

The Importance Markov chain is a new algorithm bridging the gap between ...
research
11/17/2019

Stochastic Gradient Annealed Importance Sampling for Efficient Online Marginal Likelihood Estimation

We consider estimating the marginal likelihood in settings with independ...
research
06/08/2020

Random derangements and the Ewens Sampling Formula

We study derangements of {1,2,...,n} under the Ewens distribution with p...
research
01/25/2020

The reproducing Stein kernel approach for post-hoc corrected sampling

Stein importance sampling is a widely applicable technique based on kern...
research
12/31/2019

Schrödinger Bridge Samplers

Consider a reference Markov process with initial distribution π_0 and tr...
research
07/05/2018

A multiple-try Metropolis-Hastings algorithm with tailored proposals

We present a new multiple-try Metropolis-Hastings algorithm designed to ...
research
12/08/2022

Entropy minimizing distributions are worst-case optimal importance proposals

Importance sampling of target probability distributions belonging to a g...

Please sign up or login with your details

Forgot password? Click here to reset