Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation

Résumé

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.
Fichier principal
Vignette du fichier
2312.09860.pdf (350.43 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04488225 , version 1 (04-03-2024)

Identifiants

Citer

Waïss Azizian, Guillaume Baudart, Marc Lelarge. Automatic Rao-Blackwellization for Sequential Monte Carlo with Belief Propagation. SPIGM@ICML 2023 - Structured Probabilistic Inference and Generative Modeling Workshop, Jul 2023, Honolulu, United States. ⟨hal-04488225⟩
1 Consultations
5 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More