NES spēļu spēlēšana, izmantojot imitāciju mācīšanos
Loading...
Date
Authors
Advisor
Journal Title
Journal ISSN
Volume Title
Publisher
Latvijas Universitāte
Language
lav
Abstract
Mākslīgā intelekta pielietojumi kļūst aizvien būtiskāki un izplatītāki. Pētījumā apskatīta imitāciju mācīšānās metode, kas var paātrināt mācīšanās procesu problēmu risināšanai, ja ir pieejamas eksperta demonstrācijas. Par pētījuma risināmo problēmu izvēlēta NES spēļu spēlēšana, izmantojot vispārīgu algoritmu, kas nav atkarīgs no katras spēles noteikumiem. Darbā realizēti divi imitāciju mācīšanās algoritmi, izmantojot pārraudzīto mācīšanos un stimulēto mācīšanos. Abi piegājieni beidzās neveiksmīgi.
Artificial intelligence uses are becoming more and more significant and widespread. In this paper we look at the method of imitation learning, which can speed up the learning process for problem solving, given that there is access to expert demonstrations. The problem to be solved in this paper was NES game playing using an algorithm that is not dependent on the rules of individual games. The author implemented two imitation learning algorithms using supervised learning and reinforcement learning. Both attempts ended without positive results.
Artificial intelligence uses are becoming more and more significant and widespread. In this paper we look at the method of imitation learning, which can speed up the learning process for problem solving, given that there is access to expert demonstrations. The problem to be solved in this paper was NES game playing using an algorithm that is not dependent on the rules of individual games. The author implemented two imitation learning algorithms using supervised learning and reinforcement learning. Both attempts ended without positive results.