Guardado en:
Detalles Bibliográficos
Autores principales: Liu, Ruixuan, Wang, Tianhao, Cao, Yang, Xiong, Li
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2403.09562
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909315717136384
author Liu, Ruixuan
Wang, Tianhao
Cao, Yang
Xiong, Li
author_facet Liu, Ruixuan
Wang, Tianhao
Cao, Yang
Xiong, Li
contents The pre-training and fine-tuning paradigm has demonstrated its effectiveness and has become the standard approach for tailoring language models to various tasks. Currently, community-based platforms offer easy access to various pre-trained models, as anyone can publish without strict validation processes. However, a released pre-trained model can be a privacy trap for fine-tuning datasets if it is carefully designed. In this work, we propose PreCurious framework to reveal the new attack surface where the attacker releases the pre-trained model and gets a black-box access to the final fine-tuned model. PreCurious aims to escalate the general privacy risk of both membership inference and data extraction on the fine-tuning dataset. The key intuition behind PreCurious is to manipulate the memorization stage of the pre-trained model and guide fine-tuning with a seemingly legitimate configuration. While empirical and theoretical evidence suggests that parameter-efficient and differentially private fine-tuning techniques can defend against privacy attacks on a fine-tuned model, PreCurious demonstrates the possibility of breaking up this invulnerability in a stealthy manner compared to fine-tuning on a benign pre-trained model. While DP provides some mitigation for membership inference attack, by further leveraging a sanitized dataset, PreCurious demonstrates potential vulnerabilities for targeted data extraction even under differentially private tuning with a strict privacy budget e.g. $ε=0.05$. Thus, PreCurious raises warnings for users on the potential risks of downloading pre-trained models from unknown sources, relying solely on tutorials or common-sense defenses, and releasing sanitized datasets even after perfect scrubbing.
format Preprint
id arxiv_https___arxiv_org_abs_2403_09562
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle PreCurious: How Innocent Pre-Trained Language Models Turn into Privacy Traps
Liu, Ruixuan
Wang, Tianhao
Cao, Yang
Xiong, Li
Cryptography and Security
The pre-training and fine-tuning paradigm has demonstrated its effectiveness and has become the standard approach for tailoring language models to various tasks. Currently, community-based platforms offer easy access to various pre-trained models, as anyone can publish without strict validation processes. However, a released pre-trained model can be a privacy trap for fine-tuning datasets if it is carefully designed. In this work, we propose PreCurious framework to reveal the new attack surface where the attacker releases the pre-trained model and gets a black-box access to the final fine-tuned model. PreCurious aims to escalate the general privacy risk of both membership inference and data extraction on the fine-tuning dataset. The key intuition behind PreCurious is to manipulate the memorization stage of the pre-trained model and guide fine-tuning with a seemingly legitimate configuration. While empirical and theoretical evidence suggests that parameter-efficient and differentially private fine-tuning techniques can defend against privacy attacks on a fine-tuned model, PreCurious demonstrates the possibility of breaking up this invulnerability in a stealthy manner compared to fine-tuning on a benign pre-trained model. While DP provides some mitigation for membership inference attack, by further leveraging a sanitized dataset, PreCurious demonstrates potential vulnerabilities for targeted data extraction even under differentially private tuning with a strict privacy budget e.g. $ε=0.05$. Thus, PreCurious raises warnings for users on the potential risks of downloading pre-trained models from unknown sources, relying solely on tutorials or common-sense defenses, and releasing sanitized datasets even after perfect scrubbing.
title PreCurious: How Innocent Pre-Trained Language Models Turn into Privacy Traps
topic Cryptography and Security
url https://arxiv.org/abs/2403.09562