Adversarial Attacks and Classifiers: A Review

Abstract

Through the widespread distribution of equipment (ML) for a variety of programs including medical, military, multimedia, automotive, social networking sites, and genomic there is a high risk of injury from enemy attacks (AL). The study of the opposition machine and the research center examining the production and acquisition of contradictory models, which are efforts particularly designed to betray editors, then is widely deliberate especially in the field of duplicate recognition, where small changes in image photography are made. of black boxes and methods of white box attack. Also, the study expanded the effort and examined the black box attack in contradiction of the opposition. machine learning and in-depth learning, with a strong focus on Protection from student assault. After presenting the relevant terms and objectives and a list of possible details of the attackers and opponents.