Using Interpretability of Deep Learning Neural Networks for Investigating the Effects of Adversarial Permutation