Para depositar en Docta Complutense, identifícate con tu correo @ucm.es en el SSO institucional: Haz clic en el desplegable de INICIO DE SESIÓN situado en la parte superior derecha de la pantalla. Introduce tu correo electrónico y tu contraseña de la UCM y haz clic en el botón MI CUENTA UCM, no autenticación con contraseña.
 

Variationally inferred sampling through a refined bound

Loading...
Thumbnail Image

Full text at PDC

Publication date

2021

Advisors (or tutors)

Editors

Journal Title

Journal ISSN

Volume Title

Publisher

MDPI
Citations
Google Scholar

Citation

Abstract

In this work, a framework to boost the efficiency of Bayesian inference in probabilistic models is introduced by embedding a Markov chain sampler within a variational posterior approximation. We call this framework “refined variational approximation”. Its strengths are its ease of implementation and the automatic tuning of sampler parameters, leading to a faster mixing time through automatic differentiation. Several strategies to approximate evidence lower bound (ELBO) computation are also introduced. Its efficient performance is showcased experimentally using statespace models for time-series data, a variational encoder for density estimation and a conditional variational autoencoder as a deep Bayes classifier.

Research Projects

Organizational Units

Journal Issue

Description

Unesco subjects

Keywords

Collections