Please use this identifier to cite or link to this item: http://repositorio.unicamp.br/jspui/handle/REPOSIP/108185
Type: Artigo de evento
Title: A Neural Architecture To Address Reinforcement Learning Problems
Author: De Arruda R.L.S.
Von Zuben F.J.
Abstract: In this paper, the Reinforcement Learning problem is formulated equivalently to a Markov Decision Process. We address the solution of such problem using a novel Adaptive Dynamic Programming algorithm which is based on a Multilayer Perceptron Neural Network composed of a parameterized function approximator called Wire-Fitting. Extending such established model, this work makes use of concepts of eligibility to conceive faster learning algorithms. The advantage of the proposed approach is founded on the capability to handle continuous environments and to learn a better policy while following another. Simulation results involving the automatic control of an inverted pendulum are presented to indicate the effectiveness of the proposed algorithm. © 2011 IEEE.
Editor: 
Rights: fechado
Identifier DOI: 10.1109/IJCNN.2011.6033606
Address: http://www.scopus.com/inward/record.url?eid=2-s2.0-80054769191&partnerID=40&md5=cf59c4967c0a2e6a92b285984ddc5636
Date Issue: 2011
Appears in Collections:Unicamp - Artigos e Outros Documentos

Files in This Item:
File SizeFormat 
2-s2.0-80054769191.pdf297.3 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.