A DP-based certified defense provides lower bounds on expected cumulative reward and per-state action stability for offline RL under transition- and trajectory-level poisoning, with larger certified radii than COPA.
Data Poisoning against Differentially-Private Learners: Attacks and Defenses
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Data poisoning attacks aim to manipulate the model produced by a learning algorithm by adversarially modifying the training set. We consider differential privacy as a defensive measure against this type of attack. We show that such learners are resistant to data poisoning attacks when the adversary is only able to poison a small number of items. However, this protection degrades as the adversary poisons more data. To illustrate, we design attack algorithms targeting objective and output perturbation learners, two standard approaches to differentially-private machine learning. Experiments show that our methods are effective when the attacker is allowed to poison sufficiently many training items.
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Multi-level Certified Defense Against Poisoning Attacks in Offline Reinforcement Learning
A DP-based certified defense provides lower bounds on expected cumulative reward and per-state action stability for offline RL under transition- and trajectory-level poisoning, with larger certified radii than COPA.