A Generalized Path Integral Control Approach to Reinforcement Learning

With the goal to generate more scalable algorithms with higher efficiency and fewer open parameters, reinforcement learning (RL) has recently moved towards combining classical techniques from optim...