Session Name: | Balancing Nightmares: An AI Approach to Balance Games with Overwhelming Amounts of Data |
Speaker(s): | Kazuko Manabe |
Company Name(s): | Square Enix Co., Ltd. |
Track / Format: | Programming |
Overview: | This session will share the Square Enix team's experience with the application of artificial intelligence to help improve their game balancing process for the game title, 'Grimms Notes Repage'. Specifically, speaker Kazuko Manabe will discuss a method that allowed them to balance a game with more than 10^100 choices of equipment items for battles. Kazuko is going to talk in detail about a genetic algorithm (GA) with a specific gene design that helped her team detect combinations that break game balance. In addition, she will present a data visualization method to find significant data from a large amount of log data that GA outputs. She will also explain a method to solve a problem that only appears in "player versus player", making comparisons to previous research about "player versus environment". |