Provably effective detection of effective data poisoning attacks

[Submitted on 21 Jan 2025]

View a PDF of the paper titled Provably effective detection of effective data poisoning attacks, by Jonathan Gallagher and 3 other authors

View PDF
HTML (experimental)

Abstract:This paper establishes a mathematically precise definition of dataset poisoning attack and proves that the very act of effectively poisoning a dataset ensures that the attack can be effectively detected. On top of a mathematical guarantee that dataset poisoning is identifiable by a new statistical test that we call the Conformal Separability Test, we provide experimental evidence that we can adequately detect poisoning attempts in the real world.