##阶段计划

###第一阶段

1. 麻将ENV

• support basic function
• support single patterns in win case:
  • 清一色
  • 大对
  • 七对
  • 杠牌
• support more patterns

2. 双人麻将

• dummy AI：
  • random discard an isolated tile
  • if there is a special action: do it ###第二阶段

2. 双人麻将

• ordinary AI
  • calculate the prob by every action
• smart AI
  • knows n tiles in forthcoming tilewall
  • knows part of the hand tile of the opponent ###第三阶段
• learning AI:
  • use gan in training AI and the player?
  • extend the model to four player case?
  • support more actions and patterns

###时间计划

• 1阶段： 2 月 (4月完成）
• 2阶段： 2-3 月
• 3阶段：