Primjena Monte Carlo metode i neuronskih mreža za učenje kartaških igara

undergraduate thesis

• Title (english): Application of Monte Carlo method and neural networks for learning card games
• Author: Jan Murić
• Mentor: Sonja Grgić (mentor)
• Committee member: Sonja Grgić (predsjednik povjerenstva)
• Committee member: Jelena Božek (član povjerenstva)
• Committee member: Ivana Žeger (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-05, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad istražuje primjenu Monte Carlo pretrage stabla (MCTS) i algoritma inspiriranog
AlphaZero na igru šnaps. Opisani su ključni koraci MCTS algoritma. Opisani su izazovi
implementacije MCTS-a u igri šnaps, kao što su strategije za determinizaciju nepoznatih
informacija. Također je opisan algoritam koji koristi neuronske mreže za predviđanje
politike agenta za igranje šnapsa u svrhu određivanja najboljeg poteza. Uspoređeno je pet
različitih agenata te eksperimentalni podaci pokazuju da AlphaZero inspiriran algoritam s
neuronskom mrežom ne nadmašuje performanse klasičnog MCTS algoritma u šnapsu što
upućuje na potrebu za daljnjim istraživanjem i optimizacijom.

Abstract (english)

This paper investigates the application of Monte Carlo Tree Search (MCTS) and an
AlphaZero-inspired algorithm to the game of Schnapsen. The key steps of the MCTS
algorithm are described. The challenges of implementing MCTS in Schnapsen, such as
strategies for the determinization of unknown information, are also discussed.
Additionally, an algorithm utilizing neural networks to predict the agent's policy for
playing Schnapsen in order to determine the best move is presented. Five different agents
are compared, and experimental data show that the AlphaZero-inspired algorithm with a
neural network does not outperform the classic MCTS algorithm in Schnapsen, indicating
a need for further research and optimization.

• Keywords: Monte Carlo pretraga stabla; šnaps; neuronske mreže; nepotpune igre
• Keywords (english): Monte Carlo Tree Search; Schnapsen; neural networks; imperfect information games
• Language: croatian
• URN:NBN: urn:nbn:hr:168:347846
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2024-10-12 21:39:29