Discovering Blind Spots in Reinforcement Learning

The dataset used in the paper is a collection of oracle feedback, which is used to learn a blind spot model of the target world.

BibTex: