Winner prediction in fighting games
LE3 .A278 2022
2022
Lee, Greg
Acadia University
Bachelor of Science
Honours
Computer Science
A fighting game is a video game in which two or more players battle and a winner is decided. A replay file format for a fighting game was reverse engineered to understand how important game values were stored inside saved game states. 10479 replays were gathered and the savestates of each replay were extracted into an organized dataset. The game values in each savestate were used to train an artificial neural network to predict the winner of a given match with a validation accuracy of 75%.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.
https://scholar.acadiau.ca/islandora/object/theses:3766