Inference Networks for Sequential Monte Carlo in Graphical Models

TL;DR

This paper presents a neural network-based approach to learn heuristic proposal distributions for sequential Monte Carlo in graphical models, improving inference efficiency by amortizing the inverse factorization process.

Contribution

It introduces a structured neural network that learns to approximate inverse factorizations of graphical models, enabling offline training and improved proposal distributions for SMC inference.

Findings

Neural inverse networks improve proposal quality in SMC.

The approach accelerates inference across various graphical models.

Proposal distributions are learned independently of specific datasets.

Abstract

We introduce a new approach for amortizing inference in directed graphical models by learning heuristic approximations to stochastic inverses, designed specifically for use as proposal distributions in sequential Monte Carlo methods. We describe a procedure for constructing and learning a structured neural network which represents an inverse factorization of the graphical model, resulting in a conditional density estimator that takes as input particular values of the observed random variables, and returns an approximation to the distribution of the latent variables. This recognition model can be learned offline, independent from any particular dataset, prior to performing inference. The output of these networks can be used as automatically-learned high-quality proposal distributions to accelerate sequential Monte Carlo across a diverse range of problem settings.