Multi-Objective Reinforcement Learning

VenueMODeM It's a matter of timescale: non-linear utility in successor features and multi-objective planning and learning
A visionary outlook on non-linear utility effects that unfold across multiple timescales in multi-objective reinforcement learning.
Liam P. H. Mertens, Lucas N. Alegre, Florent Delgrange, Diederik M. Roijers, Ann Nowé, Peter Vamplew