Publicación: Sum-Product Networks
Cargando...
Fecha
2019-09-24
Autores
Editor/a
Director/a
Tutor/a
Coordinador/a
Prologuista
Revisor/a
Ilustrador/a
Derechos de acceso
Atribución-NoComercial-SinDerivadas 4.0 Internacional
info:eu-repo/semantics/openAccess
info:eu-repo/semantics/openAccess
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Nacional de Educación a Distancia (España). Escuela Técnica Superior de Ingeniería Informática. Departamento de Inteligencia Artificial
Resumen
Sum-Product Networks (SPNs) are a new model that join elements from deep learning and probabilistic graphical models. They model a dataset through a hierarchical combination of mixtures and factorizations of probability distributions. Since their appearance, SPNs have obtained state of the art results in several machine learning areas. The literature about SPNs contains several obsolete papers and has neither a survey nor an introductory work addressed to the new reader. The main contribution of this work, the rst survey about SPNs, aims to ll this long-standing gap. For this work, we have thoroughly reviewed most of the existing literature. The survey presents the basic knowledge required to comprehend SPNs and then reviews how they learn their structure and parameters from data, how they perform inference, where and how they have been applied, and how they compare with other models. As an experimental contribution, the recent model of convolutional SPNs has been applied to di erent image classi cation problems and compared with convolutional neural networks.
Descripción
Categorías UNESCO
Palabras clave
Citación
Centro
Facultades y escuelas::E.T.S. de Ingeniería Informática
Departamento
Inteligencia Artificial