Continual Repeated Annealed Flow Transport Monte Carlo - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Continual Repeated Annealed Flow Transport Monte Carlo

Résumé

We propose Continual Repeated Annealed Flow Transport Monte Carlo (CRAFT), a method that combines a sequential Monte Carlo (SMC) sampler (itself a generalization of Annealed Importance Sampling) with variational inference using normalizing flows. The normalizing flows are directly trained to transport between annealing temperatures using a KL divergence for each transition. This optimization objective is itself estimated using the normalizing flow/SMC approximation. We show conceptually and using multiple empirical examples that CRAFT improves on Annealed Flow Transport Monte Carlo (Arbel et al., 2021), on which it builds and also on Markov chain Monte Carlo (MCMC) based Stochastic Normalizing Flows (Wu et al., 2020). By incorporating CRAFT within particle MCMC, we show that such learnt samplers can achieve impressively accurate results on a challenging lattice field theory example.
Fichier principal
Vignette du fichier
matthews22a.pdf (732.39 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03869105 , version 1 (24-11-2022)

Identifiants

  • HAL Id : hal-03869105 , version 1

Citer

Alexander G D G Matthews, Michael Arbel, Danilo J Rezende, Arnaud Doucet. Continual Repeated Annealed Flow Transport Monte Carlo. International Conference on Machine Learning 2022, 2022, Baltimore, United States. ⟨hal-03869105⟩
38 Consultations
17 Téléchargements

Partager

Gmail Facebook X LinkedIn More