Pseudorehearsal in actor-critic agents with neural network function approximation

Vladimir Marochko, Leonard Johard, Manuel Mazzara, Luca Longo

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Catastrophic forgetting has a significant negative impact in reinforcement learning. The purpose of this study is to investigate how pseudorehearsal can change performance of an actor-critic agent with neural-network function approximation. We tested agent in a pole balancing task and compared different pseudorehearsal approaches. We have found that pseudorehearsal can assist learning and decrease forgetting.

Original languageEnglish
Title of host publicationProceedings - 32nd IEEE International Conference on Advanced Information Networking and Applications, AINA 2018
EditorsLeonard Barolli, Tomoya Enokido, Marek R. Ogiela, Lidia Ogiela, Nadeem Javaid, Makoto Takizawa
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages644-650
Number of pages7
ISBN (Print)9781538621943
DOIs
Publication statusPublished - 2017
Externally publishedYes
Event32nd IEEE International Conference on Advanced Information Networking and Applications, AINA 2018 - Krakow, Poland
Duration: 16 May 201818 May 2018

Publication series

NameProceedings - International Conference on Advanced Information Networking and Applications, AINA
Volume2018-May
ISSN (Print)1550-445X

Conference

Conference32nd IEEE International Conference on Advanced Information Networking and Applications, AINA 2018
Country/TerritoryPoland
CityKrakow
Period16/05/1818/05/18

Keywords

  • Catastrophic forgetting
  • Neural networks
  • Pseudorehearsal
  • Reinforcement learning

Fingerprint

Dive into the research topics of 'Pseudorehearsal in actor-critic agents with neural network function approximation'. Together they form a unique fingerprint.

Cite this