Reinforcement learning based techniques in uncertain environments: problems and solutions

AlDahoul, Nouar and Htike@Muhammad Yusof, Zaw Zaw and Akmeliawati, Rini and Shafie, Amir Akramin and Khan, Sheroz (2015) Reinforcement learning based techniques in uncertain environments: problems and solutions. International Journal of Applied Engineering Research, 10 (8). pp. 20055-20066. ISSN 0973-4562

	PDF Restricted to Repository staff only Download (170kB) | Request a copy
	PDF (SCOPUS) - Published Version Restricted to Repository staff only Download (513kB) | Request a copy

Abstract

Reinforcement learning (RL) is a well-known class of machine learning algorithms used in planning and controlling of autonomous agents. Most of the issues in planning and controlling of robots are caused by uncertainties in the actuators and sensors of robots. The paper discusses important issues faced by RL in unknown and unstructured environments. It reviews problems of RL and solutions using different variants of RL namely: hierarchical RL, Bayesian model based learning, and Partially observable Markov decision processes (POMDP).

Item Type:	Article (Journal)
Additional Information:	6919/43051
Uncontrolled Keywords:	Reinforcement learning, planning, Hierarchical, Bayesian network, Partially Observable Markov decision Processes
Subjects:	A General Works > AI Indexes (General)
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Engineering
Depositing User:	Mr. Zaw Zaw Htike
Date Deposited:	28 May 2015 08:52
Last Modified:	07 Nov 2017 16:15
URI:	http://irep.iium.edu.my/id/eprint/43051

Actions (login required)

View Item