作者位次 #4/7ACL 2026Findings· Findings of ACL
SEAD : A Surrogate-free Label-only Membership Inference Attack against Pre-trained LLM s with Semantic-Aware Density
Biao Yi · Jiahao Li · Yiming Li · Yu He · Baolei Zhang · Zheli Liu · Dacheng Tao
摘要
Membership inference attacks (MIAs) aim to determine whether specific data was used to train a model. While existing MIAs against pre-trained Large Language Models (LLMs) typically require access to complete logits (probabilities), such access is sometimes unavailable in real-world deployments wh...