Enhancing the Antidote: Improved Pointwise Certifications against Poisoning Attacks

arXiv:2308.07553v2 Announce Type: replace
Abstract: Poisoning attacks can disproportionately influence model behaviour by making small changes to the training corpus. While defences against specific poisoning attacks do exist, they in general do not provide any guarantees, leaving them potentially countered by novel attacks. In contrast, by examining worst-case behaviours Certified Defences make it possible to provide guarantees of the robustness of a sample against adversarial attacks modifying a finite number of training samples, known as pointwise certification. We achieve this …

abstract arxiv attacks case certifications contrast cs.cr cs.lg general influence making novel poisoning attacks small them training type