action-interface

This paper introduces Bellman-Taylor Score Decoding, a method to handle state-dependent feasible action sets in Markov decision processes, addressing a key challenge in applying deep reinforcement learning to operations research problems.