RL-08-赵-Value函数拟合算法02-ActionValue估算03：Deep Q-learning04【DQN优化技巧②：经验回放】【Β={(s,a,r,s′)}，replay服从均匀分布】

2、技巧02：Experience replay（经验回放）问题： 什么是Experience replay? 回答：我们收集一些experience samples之后，we do NOT use these samples in the order they were collected。Instead, 我们将它们存储在一个set中，称为replay bufferB≐{ (s,a,r,s′)}\mathca