Neural networks for incremental dimensionality reduced reinforcement learning | IEEE Conference Publication | IEEE Xplore