Alpha divergence minimization in multi-class Gaussian process classification

This paper analyzes the minimization of α-divergences in the context of multi-class Gaussian process classification. For this task, several methods are explored, including memory and computationally effi cient variants of the Power Expectation Propagation algorithm, which allow for efficient trainin...

ver descrição completa

Detalhes bibliográficos
Autores: Villacampa Calvo, Carlos, Hernández Lobato, Daniel
Formato: artículo
Fecha de publicación:2020
País:España
Recursos:Universidad Autónoma de Madrid
Repositorio:Biblos-e Archivo. Repositorio Institucional de la UAM
Idioma:inglés
OAI Identifier:oai:repositorio.uam.es:10486/709938
Acesso em linha:http://hdl.handle.net/10486/709938
https://dx.doi.org/10.1016/j.neucom.2019.09.090
Access Level:acceso abierto
Palavra-chave:Gaussian processes
Expectation propagation
α-divergences
Approximate inference
Variational inference
Informática
Descrição
Resumo:This paper analyzes the minimization of α-divergences in the context of multi-class Gaussian process classification. For this task, several methods are explored, including memory and computationally effi cient variants of the Power Expectation Propagation algorithm, which allow for efficient training using stochastic gradients and mini-batches. When these methods are used for training, very large datasets (several millions of instances) can be considered. The proposed methods are also very general as they can interpolate between other popular approaches for approximate inference based on Expectation Prop agation (EP) (α →1) and Variational Bayes (VB) (α →0) simply by varying the α parameter. An exhaustive empirical evaluation analyzes the generalization properties of each of the proposed methods for differ ent values of the α parameter. The results obtained show that one can do better than EP and VB by considering intermediate values of α.