Please use this identifier to cite or link to this item:
Author(s): Gil, Paulo
Nunes, Luís
Date: 30-Jul-2013
Title: Hierarchical reinforcement learning using path clustering
Pages: Vol. I, pp. 659 - 664
Event title: Conferência Ibérica de Sistemas e Tecnologias de Informação, CISTI 2013
Keywords: reinforcement learning
Abstract: In this paper we intend to study the possibility to improve the performance of the Q-Learning algorithm, by automatically finding subgoals and making better use of the acquired knowledge. This research explores a method that allows an agent to gather information about sequences of states that lead to a goal, detect classes of common sequences and introduce the states at the end of these sequences as subgoals. We use the taxi-problem (a standard in Hierarchical Reinforcement Learning literature) and conclude that, even though this problem's scale is relatively small, in most of the cases subgoals do improve the learning speed, achieving relatively good results faster than standard Q-Learning. We propose a specific iteration interval as the most appropriate to insert subgoals in the learning process. We also found that early adoption of subgoals may lead to suboptimal learning. The extension to more challenging problems is an interesting subject for future work
Peerreviewed: Sim
Access type: Restricted Access
Appears in Collections:CTI-CRI - Comunicações a conferências internacionais

Files in This Item:
File Description SizeFormat 
Hierarchical reinforcement learning using path clustering.pdf
  Restricted Access
300,85 kBAdobe PDFView/Open Request a copy

FacebookTwitterDeliciousLinkedInDiggGoogle BookmarksMySpaceOrkut
Formato BibTex mendeley Endnote Logotipo do DeGóis Logotipo do Orcid 

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