Feed-forward Design vs. Mimic Learning for Interpretable Deep Models: A Comparative Study | IEEE Conference Publication | IEEE Xplore