Improving Adversarial robustness of deep neural networks through enforcing representation - 42Papers