Enregistré dans:
Détails bibliographiques
Auteurs principaux: Li, Chengrui, Wang, Yule, Li, Weihan, Wu, Anqi
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:https://arxiv.org/abs/2311.02516
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911769429016576
author Li, Chengrui
Wang, Yule
Li, Weihan
Wu, Anqi
author_facet Li, Chengrui
Wang, Yule
Li, Weihan
Wu, Anqi
contents Maximizing the log-likelihood is a crucial aspect of learning latent variable models, and variational inference (VI) stands as the commonly adopted method. However, VI can encounter challenges in achieving a high log-likelihood when dealing with complicated posterior distributions. In response to this limitation, we introduce a novel variational importance sampling (VIS) approach that directly estimates and maximizes the log-likelihood. VIS leverages the optimal proposal distribution, achieved by minimizing the forward $χ^2$ divergence, to enhance log-likelihood estimation. We apply VIS to various popular latent variable models, including mixture models, variational auto-encoders, and partially observable generalized linear models. Results demonstrate that our approach consistently outperforms state-of-the-art baselines, both in terms of log-likelihood and model parameter estimation.
format Preprint
id arxiv_https___arxiv_org_abs_2311_02516
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Forward $χ^2$ Divergence Based Variational Importance Sampling
Li, Chengrui
Wang, Yule
Li, Weihan
Wu, Anqi
Machine Learning
Computation
Maximizing the log-likelihood is a crucial aspect of learning latent variable models, and variational inference (VI) stands as the commonly adopted method. However, VI can encounter challenges in achieving a high log-likelihood when dealing with complicated posterior distributions. In response to this limitation, we introduce a novel variational importance sampling (VIS) approach that directly estimates and maximizes the log-likelihood. VIS leverages the optimal proposal distribution, achieved by minimizing the forward $χ^2$ divergence, to enhance log-likelihood estimation. We apply VIS to various popular latent variable models, including mixture models, variational auto-encoders, and partially observable generalized linear models. Results demonstrate that our approach consistently outperforms state-of-the-art baselines, both in terms of log-likelihood and model parameter estimation.
title Forward $χ^2$ Divergence Based Variational Importance Sampling
topic Machine Learning
Computation
url https://arxiv.org/abs/2311.02516