Domain Knowledge Alleviates Adversarial Attacks in Multi-Label Classifiers | IEEE Journals & Magazine | IEEE Xplore