Abstract: As one efficient technique in reinforcement learning, policy iteration (PI) requires an initial admissible (or stabilizing for linear systems) control policy that renders the existing ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results