Membership Inference Attacks (MIAs) determine whether a specific data point
was included in the training set of a target model. In this paper, we introduce
the Semantic Membership Inference Attack (SMIA), a novel approach that enhances
MIA performance by leveraging the semantic content of inputs and their
perturbations. SMIA trains a neural network to analyze the target model's
behavior on perturbed inputs, effectively capturing variations in output
probability distributions between members and non-members. We conduct
comprehensive evaluations on the Pythia and GPT-Neo model families using the
Wikipedia dataset. Our results show that SMIA significantly outperforms
existing MIAs; for instance, SMIA achieves an AUC-ROC of 67.39% on Pythia-12B,
compared to 58.90% by the second-best attack.

通过利用输入和其扰动的语义内容，我们引入了一种新的方法 —— 语义成员推断攻击（SMIA），从而提高成员推断攻击（MIAs）的性能。我们使用维基百科数据集对 Pythia 和 GPT-Neo 模型家族进行了全面评估，结果表明 SMIA 在 AUC-ROC 方面显著优于现有的 MIAs，例如，与第二好的攻击相比，SMIA 在 Pythia-12B 上实现了 67.39% 的 AUC-ROC，而第二好的攻击仅为 58.90%。