Differentiable Particle Filtering via Entropy-Regularized Optimal Transport
arXiv:2102.07850
Abstract
Particle Filtering (PF) methods are an established class of procedures for performing inference in non-linear state-space models. Resampling is a key ingredient of PF, necessary to obtain low variance likelihood and states estimates. However, traditional resampling methods result in PF-based loss functions being non-differentiable with respect to model and PF parameters. In a variational inference context, resampling also yields high variance gradient estimates of the PF-based evidence lower bound. By leveraging optimal transport ideas, we introduce a principled differentiable particle filter and provide convergence results. We demonstrate this novel method on a variety of applications.
arXiv comment9 pages of content + 11 pages supplementary, accepted for oral at ICML 2021
References in corpus (17)
- Auto-Encoding Variational Bayes
- A Recurrent Latent Variable Model for Sequential Data
- Fast Computation of Wasserstein Barycenters
- Near-linear time approximation algorithms for optimal transport via Sinkhorn iteration
- Interpolating between Optimal Transport and MMD using Sinkhorn Divergences
- Black box variational inference for state space models
- Wasserstein Discriminant Analysis
- Stochastic Backpropagation through Mixture Density Distributions
- On Disturbance State-Space Models and the Particle Marginal Metropolis-Hastings Sampler
- Discriminative Particle Filter Reinforcement Learning for Complex Partial Observations
- Towards Differentiable Resampling
- Linear Time Sinkhorn Divergences using Positive Features
- On embedded hidden Markov models and particle Markov chain Monte Carlo methods
- Tensor Monte Carlo: particle methods for the GPU era
- Scalable Bayesian Learning for State Space Models using Variational Inference with SMC Samplers
- Gaussian-Smooth Optimal Transport: Metric Structure and Statistical Efficiency
- End-To-End Semi-supervised Learning for Differentiable Particle Filters