Aprendizaje multitarea mediante arquitecturas neuronales basadas en subredes privadas
View/ Open
Identifiers
URI: http://hdl.handle.net/10317/702Share
Metrics
Statistics
View Usage StatisticsMetadata
Show full item recordKnowledge Area
Teoría de la Señal y las ComunicacionesSponsors
Este trabajo está parcialmente financiado por el Ministerio de Educación y Ciencia a través del proyecto TIC2002-03033.Publication date
2005-09Publisher
Universidad Politécnica de ValenciaBibliographic Citation
GARCÍA LAENCINA, Pedro José; VERDÚ MONEDERO, Rafael y SÁNCHO GÓMEZ, José Luis. Aprendizaje multitarea mediante arquitecturas neuronales basadas en subredes privadas. En: Simposium Nacional de la Unión Científica Internacional de Radio (20º: 2005: Gandía) XX Simposium Nacional de la URSI 2005. URSI 05, Gandía 14-16 de Septiembre 2005. Gandía: Universidad Politécnica de Valencia, 2005Keywords
Modelos dualesAprendizaje multitarea
Arquitecturas neuronales
Arquitectura asimétrica
Abstract
Human learning frequently involves learning several
tasks simultaneously; in particular, humans compare and
contrast similar tasks for solving a problem. Nevertheless, most
approaches to machine learning focus on the learning of a single
isolated task, Single Task Learning (STL). Most of them can be
formulated from learning several tasks related to the main task
at the same time while using a shared representation, Multitask
Learning (MTL). This type of learning improves generalization
performance for a main task by using the information contained
in other related tasks. In this article, we examine distinct schemes
used in MTL, propose an new network architecture and test each
scheme in two different problems. The proposed scheme makes
use of private subnetworks to improve the performance of MTL.
Collections
The following license files are associated with this item:
Social media