Strategic Private Audit Design (SPAD) uses a bilevel game to allocate differential privacy budgets across harm dimensions so that the welfare-weighted under-detection gap is minimized even when the audited developer responds strategically.
Proceedings of the 32nd USENIX Security Symposium , year =
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Min-K% Prob detects pretraining data in LLMs by flagging outlier low-probability words in text, achieving 7.4% better performance than prior methods on the new WIKIMIA benchmark.
citing papers explorer
-
Differentially Private Auditing Under Strategic Response
Strategic Private Audit Design (SPAD) uses a bilevel game to allocate differential privacy budgets across harm dimensions so that the welfare-weighted under-detection gap is minimized even when the audited developer responds strategically.
-
Detecting Pretraining Data from Large Language Models
Min-K% Prob detects pretraining data in LLMs by flagging outlier low-probability words in text, achieving 7.4% better performance than prior methods on the new WIKIMIA benchmark.