Dealing with Sparse Rewards in Reinforcement Learning https://arxiv.org/abs/1910.09281