University of Hradec Králové, Faculty of Informatics and Management, Department of Information Technologies
Identification and Mitigation of Threats and Vulnerabilities of Modern Machine Learning Systems
The work will examine various aspects of the harmful potential of artificial intelligence, that is, the approaches and procedures of artificial intelligence that can be the cause of attacks, and the possibilities of prevention and defence against them. Many of the possible approaches to prevent and defend against vulnerabilities from malicious AI applications must also be based on AI approaches and practices. The research will therefore be focused on both the aforementioned aspects of artificial intelligence applications and on solving typical problems of vulnerabilities in artificial intelligence systems..
Work closely with the supervisor to gather and analyze pertinent professional resources, creation of a comprehensive overview study centered around the designated subject.
Basic knowledge of AI programming is advantageous.
Systematic review, draft of a publication, software.
HU, Yupeng, et al. Artificial intelligence security: threats and countermeasures. ACM Computing Surveys (CSUR), 2021, 55.1: 1-36.