@js1037 最近在嘗試 Brain DQN相關的專案實作,作者用這個架構來玩 Flappy-Bird 主要透過Tensorflow來實作這個網絡的一些執行細節,省不少功,如果對寫遊戲AI有興趣可以參考看看 1. 文章 http://blog.csdn.net/songrotek/article/details/50951537 2. 作者專案 https://github.com/songrotek/DRL-FlappyBird