Realization and discussion of selected artificial intelligence algorithms in computer games

The study explores the usage of reinforcement learning algorithms in computer card games, such as Proximal Policy Optimization and Monte Carlo Tree Search. The aim is to evaluate the efficiency and learning ability across different scenarios, such as Blackjack and Poker Limit Hold'em. Comparat...

Full description

Saved in:
Bibliographic Details
Main Author: Yurii Tyshchenko
Format: Article
Language:English
Published: Lublin University of Technology 2025-03-01
Series:Journal of Computer Sciences Institute
Subjects:
Online Access:https://ph.pollub.pl/index.php/jcsi/article/view/6723
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The study explores the usage of reinforcement learning algorithms in computer card games, such as Proximal Policy Optimization and Monte Carlo Tree Search. The aim is to evaluate the efficiency and learning ability across different scenarios, such as Blackjack and Poker Limit Hold'em. Comparative analysis focuses on key metrics: learning speed, stability, reward evaluation and win rate. The results highlight strengths and limitations of PPO and MCTS. Also, the potential of hybrid approaches is discussed, that combine the strategic depth of MCTS with PPO's computational efficiency to create versatile AI agents capable of excelling in diverse gaming environments. The findings underscore the importance of aligning algorithmic characteristics with task specifics and domain factors.
ISSN:2544-0764