diplomsko delo
Abstract
V diplomski nalogi predstavljamo več različnih implementacij strojnega učenja računalniškega igralca za igranje igre s kartami Uno. Vsi uporabljeni algoritmi so s področja okrepitvenega učenja, saj so klasični algoritmi, ki se zanašajo na iskanje optimalne poteze na podlagi popolne informacije, neprimerni za igre z nepopolnimi informacijami. Algoritme smo primerjali glede na uspešnost v igranju proti igralcu, ki izbira naključne poteze, ter glede na krivuljo učenja, ki prikazuje pridobljeno povprečno kumulativno nagrado med procesom učenja.
Keywords
okrepitveno učenje;igra Uno;igre z nepopolnimi informacijami;igre s kartami;nevronske mreže;diplomske naloge;
Data
Language: |
Slovenian |
Year of publishing: |
2021 |
Typology: |
2.11 - Undergraduate Thesis |
Organization: |
UM FERI - Faculty of Electrical Engineering and Computer Science |
Publisher: |
[M. Praskalo] |
UDC: |
004.85(043.2) |
COBISS: |
92402947
|
Views: |
237 |
Downloads: |
17 |
Average score: |
0 (0 votes) |
Metadata: |
|
Other data
Secondary language: |
English |
Secondary title: |
Machine learning of a computer player in a card game |
Secondary abstract: |
In this graduate thesis we present several different implementations of machine learning of a computer player for playing the Uno card game. All used algorithms are from the field of reinforcement learning, as classic algorithms that rely on finding optimal moves based on complete information are unsuitable for games with incomplete information.
We compared the algorithms according to their performance when playing against a player that chooses random actions, and according to the learning curve, which represents the obtained cumulative reward during the learning process. |
Secondary keywords: |
reinforcement learning;card game Uno;imperfect information games;card games;neural networks; |
Type (COBISS): |
Bachelor thesis/paper |
Thesis comment: |
Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije |
Pages: |
IX, 31 str. |
ID: |
13329157 |