Reducing audio membership inference attack accuracy to chance: 4 defenses

Michael Lomnitz, Nina Lopatina, Paul Gamble +4

It is critical to understand the privacy and robustness vulnerabilities of machine learning models, as their implementation expands in scope. In membership inference attacks, adver…