Play Against Yourself

in an AI Rock-Paper-Scissors Game

 

#Mimic User # Machine Learning # Bayesian Algorithm # Java

​

Year 2019 (This project was explored in one of my master courses: Designing Intelligence in Interaction)

Collaborated with two of my creative teammates  Sevan Sherbetjian & Xingying Zhao

​

​

​

​

We develop an AI version of the “Rock-Paper-Scissors” game.

​

AI is mostly integrated into computer games for creating game-winning agents or improving the player’s experience. Training machine to mimic the players seems to be a less popular direction currently.

We trained the model by Bayesian algorithm to imitate the player's strategy. As a result, the player would try to beat his/her own strategy by playing "against himself/herself".

Three experiments were conducted to assess the performance of the Bayesian model.As a conclusion, the game experience can be improved by adjusting the difficulty level via k values in the model.