Guardado en:
Detalles Bibliográficos
Autores principales: Xu, Zuheng, Chen, Naitong, Campbell, Trevor
Formato: Preprint
Publicado: 2022
Materias:
Acceso en línea:https://arxiv.org/abs/2205.07475
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915316149256192
author Xu, Zuheng
Chen, Naitong
Campbell, Trevor
author_facet Xu, Zuheng
Chen, Naitong
Campbell, Trevor
contents This work presents mixed variational flows (MixFlows), a new variational family that consists of a mixture of repeated applications of a map to an initial reference distribution. First, we provide efficient algorithms for i.i.d. sampling, density evaluation, and unbiased ELBO estimation. We then show that MixFlows have MCMC-like convergence guarantees when the flow map is ergodic and measure-preserving, and provide bounds on the accumulation of error for practical implementations where the flow map is approximated. Finally, we develop an implementation of MixFlows based on uncorrected discretized Hamiltonian dynamics combined with deterministic momentum refreshment. Simulated and real data experiments show that MixFlows can provide more reliable posterior approximations than several black-box normalizing flows, as well as samples of comparable quality to those obtained from state-of-the-art MCMC methods.
format Preprint
id arxiv_https___arxiv_org_abs_2205_07475
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle MixFlows: principled variational inference via mixed flows
Xu, Zuheng
Chen, Naitong
Campbell, Trevor
Machine Learning
Computation
This work presents mixed variational flows (MixFlows), a new variational family that consists of a mixture of repeated applications of a map to an initial reference distribution. First, we provide efficient algorithms for i.i.d. sampling, density evaluation, and unbiased ELBO estimation. We then show that MixFlows have MCMC-like convergence guarantees when the flow map is ergodic and measure-preserving, and provide bounds on the accumulation of error for practical implementations where the flow map is approximated. Finally, we develop an implementation of MixFlows based on uncorrected discretized Hamiltonian dynamics combined with deterministic momentum refreshment. Simulated and real data experiments show that MixFlows can provide more reliable posterior approximations than several black-box normalizing flows, as well as samples of comparable quality to those obtained from state-of-the-art MCMC methods.
title MixFlows: principled variational inference via mixed flows
topic Machine Learning
Computation
url https://arxiv.org/abs/2205.07475