Upper Bounds on the Performance of Discretisation in Reinforcement Learning

Upper Bounds on the Performance of Discretisation in Reinforcement Learning

Reinforcement learning is a machine learning framework whereby an agent learns to perform a task by maximising its total reward received for selecting actions in each state. The policy mapping states to actions that the agent learns is either represented explicitly, or implicitly through a value fun...

Full description

Bibliographic Details
Main Author: Michael Robin Mitchley
Format: Article
Language:English
Published: South African Institute of Computer Scientists and Information Technologists 2015-12-01
Series:South African Computer Journal
Subjects:
Reinforcement learning
Tile coding
Performance Bounds
Average Case Analysis
Online Access:http://sacj.cs.uct.ac.za/index.php/sacj/article/view/284
id doaj-d287087185fe41dea2f9462369799bec
record_format Article
spelling doaj-d287087185fe41dea2f9462369799bec2020-11-25T02:42:09ZengSouth African Institute of Computer Scientists and Information TechnologistsSouth African Computer Journal1015-79992313-78352015-12-0105710.18489/sacj.v0i57.284145Upper Bounds on the Performance of Discretisation in Reinforcement LearningMichael Robin Mitchley0School of Computer Science and Applied Mathematics University of the Witwatersrand, JohannesburgReinforcement learning is a machine learning framework whereby an agent learns to perform a task by maximising its total reward received for selecting actions in each state. The policy mapping states to actions that the agent learns is either represented explicitly, or implicitly through a value function. It is common in reinforcement learning to discretise a continuous state space using tile coding or binary features. We prove an upper bound on the performance of discretisation for direct policy representation or value function approximation.http://sacj.cs.uct.ac.za/index.php/sacj/article/view/284Reinforcement learningTile codingPerformance BoundsAverage Case Analysis
collection DOAJ
language English
format Article
sources DOAJ
author Michael Robin Mitchley
spellingShingle Michael Robin Mitchley
Upper Bounds on the Performance of Discretisation in Reinforcement Learning
South African Computer Journal
Reinforcement learning
Tile coding
Performance Bounds
Average Case Analysis
author_facet Michael Robin Mitchley
author_sort Michael Robin Mitchley
title Upper Bounds on the Performance of Discretisation in Reinforcement Learning
title_short Upper Bounds on the Performance of Discretisation in Reinforcement Learning
title_full Upper Bounds on the Performance of Discretisation in Reinforcement Learning
title_fullStr Upper Bounds on the Performance of Discretisation in Reinforcement Learning
title_full_unstemmed Upper Bounds on the Performance of Discretisation in Reinforcement Learning
title_sort upper bounds on the performance of discretisation in reinforcement learning
publisher South African Institute of Computer Scientists and Information Technologists
series South African Computer Journal
issn 1015-7999
2313-7835
publishDate 2015-12-01
description Reinforcement learning is a machine learning framework whereby an agent learns to perform a task by maximising its total reward received for selecting actions in each state. The policy mapping states to actions that the agent learns is either represented explicitly, or implicitly through a value function. It is common in reinforcement learning to discretise a continuous state space using tile coding or binary features. We prove an upper bound on the performance of discretisation for direct policy representation or value function approximation.
topic Reinforcement learning
Tile coding
Performance Bounds
Average Case Analysis
url http://sacj.cs.uct.ac.za/index.php/sacj/article/view/284
work_keys_str_mv AT michaelrobinmitchley upperboundsontheperformanceofdiscretisationinreinforcementlearning
_version_ 1724775046676742144

Similar Items