You Can’t Fool All the Models: Detect Adversarial Samples via Pruning Models | IEEE Journals & Magazine | IEEE Xplore