Probabilistic guided exploration for reinforcement learning in self-organizing neural networks

Exploration is essential in reinforcement learning, which expands the search space of potential solutions to a given problem for performance evaluations. Specifically, carefully designed exploration strategy may help the agent learn faster by taking the advantage of what it has learned previously. H...

Full description

Saved in:

Bibliographic Details
Main Authors:	Wang, Peng, Zhou, Weigui Jair, Wang, Di, Tan, Ah-Hwee
Other Authors:	School of Computer Science and Engineering
Format:	Conference or Workshop Item
Language:	English
Published:	2019
Subjects:	Reinforcement Learning Self-organizing Neural Networks Engineering::Computer science and engineering
Online Access:	https://hdl.handle.net/10356/89871 http://hdl.handle.net/10220/49724
Tags:	Add Tag No Tags, Be the first to tag this record!
Institution:	Nanyang Technological University
Language:	English

id	sg-ntu-dr.10356-89871
record_format	dspace
spelling	sg-ntu-dr.10356-898712020-03-07T11:48:46Z Probabilistic guided exploration for reinforcement learning in self-organizing neural networks Wang, Peng Zhou, Weigui Jair Wang, Di Tan, Ah-Hwee School of Computer Science and Engineering 2018 IEEE International Conference on Agents (ICA) Reinforcement Learning Self-organizing Neural Networks Engineering::Computer science and engineering Exploration is essential in reinforcement learning, which expands the search space of potential solutions to a given problem for performance evaluations. Specifically, carefully designed exploration strategy may help the agent learn faster by taking the advantage of what it has learned previously. However, many reinforcement learning mechanisms still adopt simple exploration strategies, which select actions in a pure random manner among all the feasible actions. In this paper, we propose novel mechanisms to improve the existing knowledge-based exploration strategy based on a probabilistic guided approach to select actions. We conduct extensive experiments in a Minefield navigation simulator and the results show that our proposed probabilistic guided exploration approach significantly improves the convergence rate. NRF (Natl Research Foundation, S’pore) Accepted version 2019-08-21T03:56:40Z 2019-12-06T17:35:31Z 2019-08-21T03:56:40Z 2019-12-06T17:35:31Z 2018-07-01 2018 Conference Paper Wang, P., Zhou, W. J., Wang, D., & Tan, A.-H. (2018). Probabilistic guided exploration for reinforcement learning in self-organizing neural networks. 2018 IEEE International Conference on Agents (ICA). doi:10.1109/agents.2018.8460067 https://hdl.handle.net/10356/89871 http://hdl.handle.net/10220/49724 10.1109/agents.2018.8460067 209585 en © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The published version is available at: https://doi.org/10.1109/agents.2018.8460067 4 p. application/pdf
institution	Nanyang Technological University
building	NTU Library
country	Singapore
collection	DR-NTU
language	English
topic	Reinforcement Learning Self-organizing Neural Networks Engineering::Computer science and engineering
spellingShingle	Reinforcement Learning Self-organizing Neural Networks Engineering::Computer science and engineering Wang, Peng Zhou, Weigui Jair Wang, Di Tan, Ah-Hwee Probabilistic guided exploration for reinforcement learning in self-organizing neural networks
description	Exploration is essential in reinforcement learning, which expands the search space of potential solutions to a given problem for performance evaluations. Specifically, carefully designed exploration strategy may help the agent learn faster by taking the advantage of what it has learned previously. However, many reinforcement learning mechanisms still adopt simple exploration strategies, which select actions in a pure random manner among all the feasible actions. In this paper, we propose novel mechanisms to improve the existing knowledge-based exploration strategy based on a probabilistic guided approach to select actions. We conduct extensive experiments in a Minefield navigation simulator and the results show that our proposed probabilistic guided exploration approach significantly improves the convergence rate.
author2	School of Computer Science and Engineering
author_facet	School of Computer Science and Engineering Wang, Peng Zhou, Weigui Jair Wang, Di Tan, Ah-Hwee
format	Conference or Workshop Item
author	Wang, Peng Zhou, Weigui Jair Wang, Di Tan, Ah-Hwee
author_sort	Wang, Peng
title	Probabilistic guided exploration for reinforcement learning in self-organizing neural networks
title_short	Probabilistic guided exploration for reinforcement learning in self-organizing neural networks
title_full	Probabilistic guided exploration for reinforcement learning in self-organizing neural networks
title_fullStr	Probabilistic guided exploration for reinforcement learning in self-organizing neural networks
title_full_unstemmed	Probabilistic guided exploration for reinforcement learning in self-organizing neural networks
title_sort	probabilistic guided exploration for reinforcement learning in self-organizing neural networks
publishDate	2019
url	https://hdl.handle.net/10356/89871 http://hdl.handle.net/10220/49724
_version_	1681049019953971200

Probabilistic guided exploration for reinforcement learning in self-organizing neural networks

Similar Items