Carleton University - School of Computer Science Honours Project
Fall 2018
CLASSIFICATIONS AND PREDICTIONS OF A SIMPLIFIED POKER GAME
ABSTRACT
Abstract:
The problem this project is trying to solve is the ability to predict an outcome of a simplified
poker game
To solve this problem the following approach was taken – a neural network was trained to
classify poker hands and the hand odds of the opposing player were calculated by finding all the
possible hand combinations. These values were then generated as datasets for multiple poker
games and used to train a new neural network to predict a win or a loss.
Experimentation was performed on data that was downloaded from the UCI Machine Learning
Repository as well as data that was generated by the project source code. The Code
implementation of this project can be found in the following public GitHub repository
https://github.com/armankocharyan/Honours-Project/tree/master/Data%20Generator
Although the results were good, it stands that the neural network can be improved by training
it with more valid data. The predictions can also be improved by implementing other AI
methods and approaches such as decision trees and reinforcement learning.