Please use this identifier to cite or link to this item:
Title: Extracción de conocimiento de logs de póker online
Author: Antón Collado, Adrián
Director: Solé Ribalta, Albert
Tutor: Bosch Rue, Anna
Keywords: behavior analysis
deep learning
knowledge extraction
online poker
Issue Date: 5-Jan-2020
Publisher: Universitat Oberta de Catalunya (UOC)
Abstract: In recent years we are assisting to a new 'boom' of online poker. And it is not surprising since its rules are easy to learn and its dynamics simple. Beyond this simplicity it is hidden one of the toughest games in game theory. Poker online is a zero-sum, stochastic, continuous game that uses imperfect information involving up to 10 independent simultaneous players dealt around 100 hands per hour. Poker professionals use training applications based in game theory, able to calculate thousands of move combinations per second. However, due to the performance required, its use is restricted to study theorical or past hands, never online. In addition, these tools lack a part of the analysis where extra value can be extracted. Analysis on player behaviour patterns. This Master Thesis studies and analyzes a dataset of online poker hands in order to extract this kind of hidden knowledge. During the analysis an huge variety of knowledge is extracted. From knowledge that professional poker players handle perfectly, like the impact of the position in success rate, to detailed knowledge that implies deep study, like the subtle differences between betting lines. In the present Master Thesis demostrates that is possible to use Deep Learning tools can be used to analyze Texas Hold'em online Poker logs to extract knowledge and analyze the behavior of players.
Language: Spanish
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
aantoncTFM0120presentación.pdfPresentación del TFM1.41 MBAdobe PDFView/Open
aantoncTFM0120memoria.pdfMemoria del TFM3.72 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons