Overview of deep reinforcement learning in partially observable multi-agent environment of competitive online video games

Show full item record



Permalink

http://urn.fi/URN:NBN:fi:hulib-202011174484
Title: Overview of deep reinforcement learning in partially observable multi-agent environment of competitive online video games
Author: Louhio, Jaana
Contributor: University of Helsinki, Faculty of Science
Publisher: Helsingin yliopisto
Date: 2020
Language: eng
URI: http://urn.fi/URN:NBN:fi:hulib-202011174484
http://hdl.handle.net/10138/321605
Thesis level: master's thesis
Discipline: Algorithms and Machine Learning
Abstract: In the late 2010’s classical games of Go, Chess and Shogi have been considered ’solved’ by deep reinforcement learning AI agents. Competitive online video games may offer a new, more challenging environment for deep reinforcement learning and serve as a stepping stone in a path to real world applications. This thesis aims to give a short introduction to the concepts of reinforcement learning, deep networks and deep reinforcement learning. Then the thesis proceeds to look into few popular competitive online video games and to the general problems of AI development in these types of games. Deep reinforcement learning algorithms, techniques and architectures used in the development of highly competitive AI agents in Starcraft 2, Dota 2 and Quake 3 are overviewed. Finally, the results are looked into and discussed.
Subject: deep reinforcement learning
competitive online video games
POMDP
multi-agent systems


Files in this item

Total number of downloads: Loading...

Files Size Format View
Louhio_Jaana_thesis_2020.pdf 2.359Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record