ACER

2000년 도입 · 논문 12편에서 사용

ACER, or Actor Critic with Experience Replay, is an actor-critic deep reinforcement learning agent with experience replay. It can be seen as an off-policy extension of A3C, where the off-policy estimator is made feasible by: - Using Retrace Q-value estimation. - Using truncated importance sampling with bias correction. - Using a trust region policy optimization method. - Using a stochastic dueling network architecture.

출처: Sample Efficient Actor-Critic with Experience Replay

소개 논문: Sample Efficient Actor-Critic with Experience Replay