Playing Card Games on Adjustable Shuffling with Reinforcement Learning Agents

Ryotaro Suzuki, Naoki Fukuta

Abstract: This paper shows an example application system that supports the users to find a suitable customized shuffling method for playing card games. To enable the users to utilize the characteristics of each shuffling method in each game playing, we develop a prototype gaming platform with a shuffling adjustment support mechanism that assists players in selecting the preferred shuffling method and the custom number of shuffling times to fit to their playing styles. We also develop a prototype player agent that behaves as the good counterpart in the exciting gameplay defined by Isabella et al. in 2004, applying QS-learning technique to allow the agent to change its behavior according to the characteristics of the shuffling.