Dynamic Bayesian Networks for State- and Action-Space Modelling in Reinforcement Learning

Леховицький, Д. І.; Ховрат, А. В.

Публікація:
Dynamic Bayesian Networks for State- and Action-Space Modelling in Reinforcement Learning

dc.contributor.author	Леховицький, Д. І.
dc.contributor.author	Ховрат, А. В.
dc.date.accessioned	2018-06-05T12:16:44Z
dc.date.available	2018-06-05T12:16:44Z
dc.date.issued	2018
dc.description.abstract	In recent years Reinforcement Learning has proven its efficiency in solving problems of sequential decision making, formalized with a concept called Markov Decision Process. Though, there is a lot of problems: high computational complexity for multivariate state- and action-space problems, needs to handle missing data and hidden variables, lack of both good model and a sufficient number of episodes for constructing an optimal policy. In this work we suggest Dynamic Bayesian networks (DBNs) as a solution. These models provide an elegant and compact representation of joint state-action space, efficient inference algorithms, which include Monte-Carlo methods and Belief Propagation, and can be used in Dyna-Q Algorithm for integrating real-world and simulated experience.	uk_UA
dc.identifier.citation	Lekhovitsky D., Khovrat A. Dynamic Bayesian Networks for State- and Action-Space Modelling in Reinforcement Learning / D. Lekhovitsky, A. Khovrat // Радіоелектроніка та молодь у XXI столітті : матеріали 22-го Міжнар. молодіжного форуму, 17–19 апр. 2018 г. – Харків : ХНУРЕ, 2018. – С. 118–119.	uk_UA
dc.identifier.uri	http://openarchive.nure.ua/handle/document/5806
dc.language.iso	en	uk_UA
dc.publisher	ХНУРЕ	uk_UA
dc.subject	Markov Decision Process	uk_UA
dc.subject	Dynamic Bayesian networks	uk_UA
dc.subject	Reinforcement Learning	uk_UA
dc.title	Dynamic Bayesian Networks for State- and Action-Space Modelling in Reinforcement Learning	uk_UA
dc.type	Thesis	uk_UA
dspace.entity.type	Publication