Detecting Adversarial Image Examples in Deep Neural Networks with Adaptive Noise Reduction
Detecting Adversarial Image Examples in Deep Neural Networks with Adaptive Noise Reduction
Recently, many studies have demonstrated deep neural network (DNN) classifiers can be fooled by the adversarial example, which is crafted via introducing some perturbations into an original sample. Accordingly, some powerful defense techniques were proposed. However, existing defense techniques often require modifying the target model or depend on the prior …