Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network

Link to publisher's homepage at http://link.springer.com/

Saved in:

Bibliographic Details
Main Authors:	Mohamad Faizal, Samsudin, Sawatsubashi, Yoshito, Katsunari, Shibata
Other Authors:	faizalsamsudin@unimap.edu.my
Format:	Article
Language:	English
Published:	Springer-Verlag 2014
Subjects:	Multi-step discrete state transition Recurrent neural network Reinforcement learning
Online Access:	http://dspace.unimap.edu.my/xmlui/handle/123456789/30948
Tags:	Add Tag No Tags, Be the first to tag this record!
Institution:	Universiti Malaysia Perlis
Language:	English

id	my.unimap-30948
record_format	dspace
spelling	my.unimap-309482014-01-10T08:55:22Z Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network Mohamad Faizal, Samsudin Sawatsubashi, Yoshito Katsunari, Shibata faizalsamsudin@unimap.edu.my bashis8@yahoo.co.jp shibata@oita-u.ac.jp Multi-step discrete state transition Recurrent neural network Reinforcement learning Link to publisher's homepage at http://link.springer.com/ For developing a robot that learns long and complicated action sequences in the real-world, autonomous learning of multi-step discrete state transition is significant. To realize the multi-step discrete state transition in a neural network is generally thought to be difficult because of basically the needs to hold the state while performing the transition between the states when needed. In this paper, only through the reinforcement learning using rewards and punishments in a simple learning system consisting of a recurrent neural network (RNN), it is shown that a multi-step discrete state transition emerged through learning in a continuous state-action space. It is shown that in a two-switch task, two states transition represented by two types of hidden nodes emerged through the learning. In addition, it is shown that the contribution of the dynamics by the interaction between the RNN and the environment based on the discrete state transitions leads to repetition of the interesting behavior when no reward is given at the goal. 2014-01-02T08:46:59Z 2014-01-02T08:46:59Z 2012 Article Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7664 LNCS(Part2), 2012, pages 583-590 978-364234480-0 0302-9743 http://link.springer.com/chapter/10.1007%2F978-3-642-34481-7_71 http://hdl.handle.net/123456789/30948 en Springer-Verlag
institution	Universiti Malaysia Perlis
building	UniMAP Library
collection	Institutional Repository
continent	Asia
country	Malaysia
content_provider	Universiti Malaysia Perlis
content_source	UniMAP Library Digital Repository
url_provider	http://dspace.unimap.edu.my/
language	English
topic	Multi-step discrete state transition Recurrent neural network Reinforcement learning
spellingShingle	Multi-step discrete state transition Recurrent neural network Reinforcement learning Mohamad Faizal, Samsudin Sawatsubashi, Yoshito Katsunari, Shibata Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
description	Link to publisher's homepage at http://link.springer.com/
author2	faizalsamsudin@unimap.edu.my
author_facet	faizalsamsudin@unimap.edu.my Mohamad Faizal, Samsudin Sawatsubashi, Yoshito Katsunari, Shibata
format	Article
author	Mohamad Faizal, Samsudin Sawatsubashi, Yoshito Katsunari, Shibata
author_sort	Mohamad Faizal, Samsudin
title	Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
title_short	Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
title_full	Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
title_fullStr	Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
title_full_unstemmed	Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
title_sort	emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network
publisher	Springer-Verlag
publishDate	2014
url	http://dspace.unimap.edu.my/xmlui/handle/123456789/30948
_version_	1643796421056397312

Emergence of multi-step discrete state transition through reinforcement learning with a recurrent neural network

Similar Items