Abstract
We present a novel method for learning a set of disentangled reward functions that sum to the original environment reward and are constrained to be independently achievable. We define independent achievability in terms of value functions with respect to achieving one learned reward while pursuing another learned reward. Empirically, we illustrate that our method can learn meaningful reward decompositions in a variety of domains and that these decompositions exhibit some form of generalization performance when the environment’s reward is modified. Theoretically, we derive results about the effect of maximizing our method’s objective on the resulting reward functions and their corresponding optimal policies.
Abstract (translated by Google)
URL
http://arxiv.org/abs/1901.08649