Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation

Abstract

Exact Bayesian inference on state-space models (SSM) is in general untractable and, unfortunately, basic Sequential Monte Carlo (SMC) methods do not yield correct approximations for complex models. In this paper, we propose a mixed inference algorithm that computes closed-form solutions using Belief Propagation as much as possible, and falls back to sampling-based SMC methods when exact computations fail. This algorithm thus implements automatic Rao-Blackwellization and is even exact for Gaussian tree models.

Cite

Text

Azizian et al. "Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation." ICML 2023 Workshops: SPIGM, 2023.

Markdown

[Azizian et al. "Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation." ICML 2023 Workshops: SPIGM, 2023.](https://mlanthology.org/icmlw/2023/azizian2023icmlw-automatic/)

BibTeX

@inproceedings{azizian2023icmlw-automatic,
  title     = {{Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation}},
  author    = {Azizian, Waïss and Baudart, Guillaume and Lelarge, Marc},
  booktitle = {ICML 2023 Workshops: SPIGM},
  year      = {2023},
  url       = {https://mlanthology.org/icmlw/2023/azizian2023icmlw-automatic/}
}