Background Adversarial purification is a defense mechanism for safe-guarding classifiers against adversarial attacks without knowing the type of attacks or training of the classifier. These techniques analyze and eliminate adversarial perturbations from the attacked inputs, and help to restore purified samples that retain similarity to the attacked...

Background Social media users generate a tremendous amount of data in forms such as profile information, network connections, and online posts. Online vendors use this data to understand user preferences and further predict their future needs. However, because user-generated data is rich in content, the data can be used by malicious attackers to infer...