DRN: Bringing Greedy Layer-Wise Training into Time Dimension | IEEE Conference Publication | IEEE Xplore