Adversarial Attack Detection Based on Example Semantics and Model Activation Features | IEEE Conference Publication | IEEE Xplore