Open Access System for Information Sharing

Login Library

 

Article
Cited 10 time in webofscience Cited 15 time in scopus
Metadata Downloads
Full metadata record
Files in This Item:
There are no files associated with this item.
DC FieldValueLanguage
dc.contributor.authorBAEK, JONGCHAN-
dc.contributor.authorJEON, HAYEONG-
dc.contributor.authorPARK, JONGYEOK-
dc.contributor.authorLEE, HAKJUN-
dc.contributor.authorHAN, SOOHEE-
dc.date.accessioned2021-06-01T01:53:08Z-
dc.date.available2021-06-01T01:53:08Z-
dc.date.created2021-03-08-
dc.date.issued2021-09-
dc.identifier.issn0278-0046-
dc.identifier.urihttps://oasis.postech.ac.kr/handle/2014.oak/105104-
dc.description.abstractRecent advancements in deep reinforcement learning for real control tasks have received interest from many researchers and field engineers in a variety of industrial areas. However, in most cases, optimal policies obtained by deep reinforcement learning are difficult to implement on cost-effective and lightweight platforms such as mobile devices. This can be attributed to their computational complexity and excessive memory usage. For this reason, this study proposes an off-policy deep reinforcement learning algorithm called the sparse variational deterministic policy gradient (SVDPG). SVDPG provides highly efficient policy network compression under the standard reinforcement learning framework. The proposed SVDPG integrates Bayesian pruning, which is known as a state-of-the-art neural network compression technique, with the policy update in an actor-critic architecture for reinforcement learning. It is demonstrated that SVDPG achieves a high compression rate of policy networks for continuous control benchmark tasks while preserving a competitive performance. The superiority of SVDPG in low-computing power devices is proven by comparing the level of compression in terms of the memory requirements and computation time on a commercial microcontroller unit. Finally, it is confirmed that the proposed SVDPG is also reliable in real-world scenarios since it can be applied to the swing-up control of an inverted pendulum system.-
dc.languageEnglish-
dc.publisherInstitute of Electrical and Electronics Engineers-
dc.relation.isPartOfIEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS-
dc.subjectBenchmarking-
dc.subjectCost effectiveness-
dc.subjectDeep learning-
dc.subjectReal time control-
dc.subjectReinforcement learning-
dc.subjectActor-critic architectures-
dc.subjectCompetitive performance-
dc.subjectContinuous control-
dc.subjectHigh compressions-
dc.subjectInverted pendulum system-
dc.subjectMemory requirements-
dc.subjectMicrocontroller unit-
dc.subjectReal-world scenario-
dc.subjectLearning algorithms-
dc.titleSparse Variational Deterministic Policy Gradient for Continuous Real Time Control-
dc.typeArticle-
dc.identifier.doi10.1109/TIE.2020.3021607-
dc.type.rimsART-
dc.identifier.bibliographicCitationIEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, v.68, no.10, pp.9800 - 9810-
dc.identifier.wosid000670541800070-
dc.citation.endPage9810-
dc.citation.number10-
dc.citation.startPage9800-
dc.citation.titleIEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS-
dc.citation.volume68-
dc.contributor.affiliatedAuthorBAEK, JONGCHAN-
dc.contributor.affiliatedAuthorJEON, HAYEONG-
dc.contributor.affiliatedAuthorPARK, JONGYEOK-
dc.contributor.affiliatedAuthorLEE, HAKJUN-
dc.contributor.affiliatedAuthorHAN, SOOHEE-
dc.identifier.scopusid2-s2.0-85112513702-
dc.description.journalClass1-
dc.description.journalClass1-
dc.description.isOpenAccessN-
dc.type.docTypeArticle-
dc.subject.keywordAuthorBayes methods-
dc.subject.keywordAuthorMachine learning-
dc.subject.keywordAuthorNeural networks-
dc.subject.keywordAuthorLearning (artificial intelligence)-
dc.subject.keywordAuthorComputational modeling-
dc.subject.keywordAuthorStandards-
dc.subject.keywordAuthorOptimization-
dc.subject.keywordAuthorBayesian compression-
dc.subject.keywordAuthordeep reinforcement learning-
dc.subject.keywordAuthorinverted pendulum system-
dc.subject.keywordAuthorsparse Bayesian deep learning-
dc.relation.journalWebOfScienceCategoryAutomation & Control Systems-
dc.relation.journalWebOfScienceCategoryEngineering, Electrical & Electronic-
dc.relation.journalWebOfScienceCategoryInstruments & Instrumentation-
dc.description.journalRegisteredClassscie-
dc.description.journalRegisteredClassscopus-
dc.relation.journalResearchAreaAutomation & Control Systems-
dc.relation.journalResearchAreaEngineering-
dc.relation.journalResearchAreaInstruments & Instrumentation-

qr_code

  • mendeley

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Related Researcher

Researcher

한수희HAN, SOOHEE
Dept of Electrical Enginrg
Read more

Views & Downloads

Browse