PROTECTION AGAINST ATTACKS ON MACHINE LEARNING SYSTEMS ON THE EXAMPLE OF EVADIATION ATTACKS IN MEDICAL IMAGE ANALYSIS

This paper is about the adversarial attacks on machine learning systems that analyze
medical images. The authors review the existing attacks, conducts their systematization and practical
feasibility. The article contains an analysis of existing methods of protection against adversarial
attacks on machine learning systems. It describes the peculiarities of medical images. The
authors solve the problem of protection against adversarial attacks for these images based on several
defensive methods. The authors have determined the most relevant protection methods, their
implementation and testing on practical examples — the analysis of COVID‑19 patient’s images