Adversarial Detector with Robust Classifier
This addresses security vulnerabilities in AI systems for applications like image recognition, but it appears incremental as it builds on existing detector and robust classifier concepts.
The paper tackles the problem of adversarial examples causing misclassification in deep neural networks by proposing a novel adversarial detector that combines a robust classifier with a plain one, achieving superior performance over a state-of-the-art detector without a robust classifier.
Deep neural network (DNN) models are wellknown to easily misclassify prediction results by using input images with small perturbations, called adversarial examples. In this paper, we propose a novel adversarial detector, which consists of a robust classifier and a plain one, to highly detect adversarial examples. The proposed adversarial detector is carried out in accordance with the logits of plain and robust classifiers. In an experiment, the proposed detector is demonstrated to outperform a state-of-the-art detector without any robust classifier.