【论文阅读】QUEEN: Query Unlearning against Model Extraction(2024)
摘要Modelextractionattacks(模型提取攻击)currentlyposeanon-negligiblethreat(不可忽视的威胁)tothesecurity(安全性)andprivacy(隐私性)ofdeeplearningmodels.Byqueryingthemodelwithasmalldataset(通过小数据集查询模型)andusingthequeryresultsa