Amortized Rejection Sampling in Universal Probabilistic Programming

Autor:	Naderiparizi, S, Scibior, A, Munk, A, Ghadiri, M, Güneş Baydin, A, Gram-Hansen, B, Schroeder de Witt, C, Zinkov, R, Torr, PHS, Rainforth, T, Teh, YW, Wood, F
Jazyk:	angličtina
Rok vydání:	2019
Předmět:	FOS: Computer and information sciences Computer Science - Machine Learning Artificial Intelligence (cs.AI) Computer Science - Artificial Intelligence Statistics - Machine Learning Machine Learning (stat.ML) Machine Learning (cs.LG)
Popis:	Naive approaches to amortized inference in probabilistic programs with unbounded loops can produce estimators with infinite variance. This is particularly true of importance sampling inference in programs that explicitly include rejection sampling as part of the user-programmed generative procedure. In this paper we develop a new and efficient amortized importance sampling estimator. We prove finite variance of our estimator and empirically demonstrate our method's correctness and efficiency compared to existing alternatives on generative programs containing rejection sampling loops and discuss how to implement our method in a generic probabilistic programming framework. Comment: AISTATS 2022 camera ready
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_dedup___::a127fed03257ddb282757dd938aa69a0 http://arxiv.org/abs/1910.09056 Zobrazit plný text záznamu