CS294-112 深度强化学习 秋季学期(伯克利)NO.9 Learning policies by imitating optimal controllers
make compromise between learnt policy and minimal cost!
π hat is using states
π theta is using observations