<ul><li>Model Ownership Resolution (MOR) in ML security addresses the challenge of proving ownership of sophisticated ML models developed with substantial resources.</li><li>Defending against multiple types of attacks and privacy breaches simultaneously is a key challenge as attackers evolve and employ various techniques, creating an interconnected threat landscape.</li><li>ML security and privacy demand a strategic and comprehensive approach, emphasizing a holistic understanding of threats and defenses to address evolving challenges.</li><li>The discussion by Professor Asokan underscores the complexity of securing and ensuring privacy in ML systems, highlighting the need to anticipate future threats and understand the detailed interactions between threats and defenses.</li></ul>

Talk on “Meta Concerns in ML Security/Privacy” with Professor N. Asokan from University of Waterloo

Discover more