Global Reinforcement Learning : Beyond Linear and Convex Rewards via Submodular Semi-gradient Methods.

Riccardo De Santi, Manish Prajapat, Andreas Krause

Browse the full ICML paper archive.