In reinforcement learning (RL), agents often operate in partially observed
and uncertain environments. Model-based RL suggests that this is best achieved
by learning and exploiting a probabilistic model of the wo
该文章介绍了 Active Inference 的理论,探讨了将行动和规划转化为一个贝叶斯推理问题以最小化可变自由能的方法。 它提出了一种新颖的深度 Active Inference 算法,该算法通过使用深度神经网络作为灵活的函数逼近器来逼近关键密度,从而使 Active Inference 能够处理更大更复杂的任务,并展示了与强化学习的有趣关联。