Goal-driven dimensionality reduction for reinforcement learning | IEEE Conference Publication | IEEE Xplore