How to penalize the decision boundary in Deep neural networks?

Adversarial Training squeeze the decision boundary. how to penalty it? I want to increase the boundary gap between the two classes.

30 June 2020 7,725 0 View