Particle Smoothing Variational Objectives

Abstract

A body of recent work has focused on constructing a variational family of filtered distributions using Sequential Monte Carlo (SMC). Inspired by this work, we introduce Particle Smoothing Variational Objectives (SVO), a novel backward simulation technique and smoothed approximate posterior defined through a subsampling process. SVO augments support of the proposal and boosts particle diversity. Recent literature argues that increasing the number of samples K to obtain tighter variational bounds may hurt the proposal learning, due to a signal-to-noise ratio (SNR) of gradient estimators decreasing at the rate $\mathcal{O}(1/\sqrt{K})$. As a second contribution, we develop theoretical and empirical analysis of the SNR in filtering SMC, which motivates our choice of biased gradient estimators. We prove that introducing bias by dropping Categorical terms from the gradient estimate or using Gumbel-Softmax mitigates the adverse effect on the SNR. We apply SVO to three nonlinear latent dynamics tasks and provide statistics to rigorously quantify the predictions of filtered and smoothed objectives. SVO consistently outperforms filtered objectives when given fewer Monte Carlo samples on three nonlinear systems of increasing complexity.

13 pages, 5 figures

References in corpus (8)

• Auto-Encoding Variational Bayes
• Categorical Reparameterization with Gumbel-Softmax
• The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
• Deep Kalman Filters
• Black box variational inference for state space models
• Auto-Encoding Sequential Monte Carlo
• Importance Weighting and Variational Inference
• Graphical model inference: Sequential Monte Carlo meets deterministic approximations

Cited by in corpus (5)

• Markovian Score Climbing: Variational Inference with KL(p||q)
• Nonlinear Evolution via Spatially-Dependent Linear Dynamics for Electrophysiology and Calcium Data
• Variational Marginal Particle Filters
• Ensemble Kalman Variational Objectives: Nonlinear Latent Trajectory Inference with A Hybrid of Variational Inference and Ensemble Kalman Filter
• Variational Combinatorial Sequential Monte Carlo Methods for Bayesian Phylogenetic Inference