Amplifying Training Data Exposure through Fine-Tuning with Pseudo-Labeled Memberships

arXiv:2402.12189v1 Announce Type: cross
Abstract: Neural language models (LMs) are vulnerable to training data extraction attacks due to data memorization. This paper introduces a novel attack scenario wherein an attacker adversarially fine-tunes pre-trained LMs to amplify the exposure of the original training data. This strategy differs from prior studies by aiming to intensify the LM's retention of its pre-training dataset. To achieve this, the attacker needs to collect generated texts that are closely aligned with the pre-training data. However, without …

abstract amplify arxiv attacks cs.cl cs.cr cs.lg data data extraction extraction fine-tuning language language models lms novel paper prior strategy studies through training training data type vulnerable