20.500.12556/DKUM-79581 Implementacija inteligentnega agenta za igro lažnivec na osnovi statističnega modela diplomsko delo Implementation of intelligent agent for the Cheat card game based on statistical model Lažnivec je popularna igra s kartami za dva igralca, ki si med seboj razdelita karte in jih izmenično odlagata na kup. Pri sprejemanju svojih odločitev se odločata glede na število kart v rokah in na kupu ter glede na predvidevanje nasprotnikovih potez. V tem diplomskem delu smo predstavili lastno implementacijo igre, ki vključuje več vrst računalniških agentov, definiranih na osnovi statističnih modelov. Poleg tradicionalnih hevristik igranja smo za njihovo definicijo uporabili tudi genetski algoritem, kjer se uspešnost igranja agentov preverja s turnirjem. Slednjega smo izvedli po principu, da vsak igra z vsakim, in z igranjem agentov proti človeškemu nasprotniku. Rezultati igre med agenti so pokazali, da za igro lažnivec ne obstaja splošno najuspešnejša strategija, ki bi zagotavljala zmago v večini primerov. Rezultati igre proti človeškemu igralcu pa so razkrili ključno slabost statistično podprtih modelov, ki se niso zmožni prilagoditi spremembam v nasprotnikovi strategiji. Cheat is a popular card game, where two players split a set of cards amongst each other and, one by one, place them on a pile. They base their decisions on the number of cards in their hands and on the pile, while also trying to anticipate their opponent’s moves. In this diploma thesis, we present a software implementation of the game, together with several types of computer agents, defined by statistical models. In addition to traditional heuristics, their definition was achieved using genetic algorithm, where success of the agents was assessed with a tournament amongst them as well as against a human opponent. The results of the games between the agents showed that, in general, there is no most successful strategy that would guarantee victory in most cases. On the other hand, the results of the game against the human opponent revealed a key weakness of statistical models, that is their inability to adapt to the changes in the opponent’s strategy. lažnivec igra s kartami genetski algoritem umetna inteligenca računalniški agenti Cheat Card game Genetic algorithm Artificial Intelligence Software agents true true false [D. Mikek] Slovenski jezik Angleški jezik Diplomsko delo/naloga 2021-07-24 15:32:44 2021-10-18 07:32:49 2022-08-09 21:45:12 0000-00-00 00:00:00 2021 0 Maribor 0 Maribor VI, 25 str. 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.8:794.4(043.2) 89640195 UN_Mikek_David_2021.pdf UN_Mikek_David_2021.pdf 1 2F14BAA1B2D3331879301BFE017A7686 ce2aa8820add4d27b55766f1d27e0e3910b3c5871291a8f3f1b202e0c357b867 7417f772-f42c-11eb-accf-00155d581298 20.500.12556/dkum/c4286b80-a9e5-46de-a169-fef2521d869e https://dk.um.si/Dokument.php?lang=slv&id=151494 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0