Transfer learning through policy abstraction using learning vector quantization

Reinforcement learning (RL) enables an agent to find a solution to a problem by interacting with the environment. However, the learning process always starts from scratch and possibly takes a long time. Here, knowledge transfer between tasks is considered. In this paper, we argue that an abstraction...

Full description

Bibliographic Details
Main Authors:	Ahmad Afif, Mohd Faudzi, Takano, Hirotaka, Murata, Jun'ichi
Format:	Article
Language:	English
Published:	UTeM 2018
Subjects:	TK Electrical engineering. Electronics Nuclear engineering
Online Access:	http://umpir.ump.edu.my/id/eprint/21033/ http://umpir.ump.edu.my/id/eprint/21033/ http://umpir.ump.edu.my/id/eprint/21033/1/Transfer%20learning%20through%20policy%20abstraction.pdf

id	ump-21033
recordtype	eprints
spelling	ump-210332018-08-20T07:10:21Z http://umpir.ump.edu.my/id/eprint/21033/ Transfer learning through policy abstraction using learning vector quantization Ahmad Afif, Mohd Faudzi Takano, Hirotaka Murata, Jun'ichi TK Electrical engineering. Electronics Nuclear engineering Reinforcement learning (RL) enables an agent to find a solution to a problem by interacting with the environment. However, the learning process always starts from scratch and possibly takes a long time. Here, knowledge transfer between tasks is considered. In this paper, we argue that an abstraction can improve the transfer learning. Modified learning vector quantization (LVQ) that can manipulate its network weights is proposed to perform an abstraction, an adaptation and a precaution. At first, the abstraction is performed by extracting an abstract policy out of a learned policy which is acquired through conventional RL method, Q-learning. The abstract policy then is used in a new task as prior information. Here, the adaptation or policy learning as well as new task's abstract policy generating are performed using only a single operation. Simulation results show that the representation of acquired abstract policy is interpretable, that the modified LVQ successfully performs policy learning as well as generates abstract policy and that the application of generalized common abstract policy produces better results by more effectively guiding the agent when learning a new task. UTeM 2018 Article PeerReviewed pdf en cc_by http://umpir.ump.edu.my/id/eprint/21033/1/Transfer%20learning%20through%20policy%20abstraction.pdf Ahmad Afif, Mohd Faudzi and Takano, Hirotaka and Murata, Jun'ichi (2018) Transfer learning through policy abstraction using learning vector quantization. Journal of Telecommunication, Electronic and Computer Engineering, 10 (1-3). pp. 163-168. ISSN 2180-1843 (Print); 2289-8131 (Online) http://journal.utem.edu.my/index.php/jtec/article/view/3505/2453
repository_type	Digital Repository
institution_category	Local University
institution	Universiti Malaysia Pahang
building	UMP Institutional Repository
collection	Online Access
language	English
topic	TK Electrical engineering. Electronics Nuclear engineering
spellingShingle	TK Electrical engineering. Electronics Nuclear engineering Ahmad Afif, Mohd Faudzi Takano, Hirotaka Murata, Jun'ichi Transfer learning through policy abstraction using learning vector quantization
description	Reinforcement learning (RL) enables an agent to find a solution to a problem by interacting with the environment. However, the learning process always starts from scratch and possibly takes a long time. Here, knowledge transfer between tasks is considered. In this paper, we argue that an abstraction can improve the transfer learning. Modified learning vector quantization (LVQ) that can manipulate its network weights is proposed to perform an abstraction, an adaptation and a precaution. At first, the abstraction is performed by extracting an abstract policy out of a learned policy which is acquired through conventional RL method, Q-learning. The abstract policy then is used in a new task as prior information. Here, the adaptation or policy learning as well as new task's abstract policy generating are performed using only a single operation. Simulation results show that the representation of acquired abstract policy is interpretable, that the modified LVQ successfully performs policy learning as well as generates abstract policy and that the application of generalized common abstract policy produces better results by more effectively guiding the agent when learning a new task.
format	Article
author	Ahmad Afif, Mohd Faudzi Takano, Hirotaka Murata, Jun'ichi
author_facet	Ahmad Afif, Mohd Faudzi Takano, Hirotaka Murata, Jun'ichi
author_sort	Ahmad Afif, Mohd Faudzi
title	Transfer learning through policy abstraction using learning vector quantization
title_short	Transfer learning through policy abstraction using learning vector quantization
title_full	Transfer learning through policy abstraction using learning vector quantization
title_fullStr	Transfer learning through policy abstraction using learning vector quantization
title_full_unstemmed	Transfer learning through policy abstraction using learning vector quantization
title_sort	transfer learning through policy abstraction using learning vector quantization
publisher	UTeM
publishDate	2018
url	http://umpir.ump.edu.my/id/eprint/21033/ http://umpir.ump.edu.my/id/eprint/21033/ http://umpir.ump.edu.my/id/eprint/21033/1/Transfer%20learning%20through%20policy%20abstraction.pdf
first_indexed	2023-09-18T22:30:42Z
last_indexed	2023-09-18T22:30:42Z
_version_	1777416266067214336

Transfer learning through policy abstraction using learning vector quantization

Similar Items