It seems neural nets are quite prone to this sort of overfitting, given the article on adversarial objects from the other day: http://www.labsix.org/physical-ob

It seems neural nets are quite prone to this sort of overfitting, given the article on adversarial objects from the other day: http://www.labsix.org/physical-objects-that-fool-neural-nets/ http://www.labsix.org/physical-objects-that-fool-neural-nets...