Unrolling Particles: Unsupervised Learning of Sampling Distributions

TL;DR

This paper introduces an unsupervised learning approach to optimize sampling distributions in particle filtering, using neural networks to improve estimation accuracy in complex systems.

Contribution

It proposes a novel method to learn sampling distributions via algorithm unrolling, reducing degeneracy and enhancing particle filter performance.

Findings

Learned distributions reduce sample degeneracy.

Improved estimates across diverse scenarios.

Neural network-based approach outperforms traditional methods.

Abstract

Particle filtering is used to compute good nonlinear estimates of complex systems. It samples trajectories from a chosen distribution and computes the estimate as a weighted average. Easy-to-sample distributions often lead to degenerate samples where only one trajectory carries all the weight, negatively affecting the resulting performance of the estimate. While much research has been done on the design of appropriate sampling distributions that would lead to controlled degeneracy, in this paper our objective is to \emph{learn} sampling distributions. Leveraging the framework of algorithm unrolling, we model the sampling distribution as a multivariate normal, and we use neural networks to learn both the mean and the covariance. We carry out unsupervised training of the model to minimize weight degeneracy, relying only on the observed measurements of the system. We show in simulations that the resulting particle filter yields good estimates in a wide range of scenarios.