recurrent-networks

This paper introduces Neural Co-state Policies, establishing a formal link between recurrent reinforcement learning hidden states and the Pontryagin minimum principle to improve interpretability and robustness.