New Perspective of Interpretability of Deep Neural Networks | IEEE Conference Publication | IEEE Xplore