Constructive Noise Defeats Adversarial Noise: Adversarial Example Detection for Commercial DNN Services

—Commercial DNN services have been developed in the form of machine learning as a service (MLaaS). To mitigate the potential threats of adversarial examples, various detection methods have been proposed. However, the existing methods usually require access to details or the training dataset of the target model, which is commonly unavailable in MLaaS scenarios. Their detection accuracy experiences a significant d rop i n a setting where neither the details nor the training dataset of the target model can be acquired.