SecretGen: Privacy Recovery on Pre-Trained Models via Distribution Discrimination

TL;DR

This paper introduces SecretGen, a novel framework for recovering private training data from pre-trained models without prior knowledge, highlighting privacy vulnerabilities and providing benchmarks for privacy assessment.

Contribution

SecretGen is a new method that effectively reconstructs private data from pre-trained models without needing true class predictions, advancing privacy attack techniques.

Findings

SecretGen can recover private data without prior true class information.

It outperforms baseline methods when prior knowledge is available.

The study provides metrics to quantify privacy risks in pre-trained models.

Abstract

Transfer learning through the use of pre-trained models has become a growing trend for the machine learning community. Consequently, numerous pre-trained models are released online to facilitate further research. However, it raises extensive concerns on whether these pre-trained models would leak privacy-sensitive information of their training data. Thus, in this work, we aim to answer the following questions: "Can we effectively recover private information from these pre-trained models? What are the sufficient conditions to retrieve such sensitive information?" We first explore different statistical information which can discriminate the private training distribution from other distributions. Based on our observations, we propose a novel private data reconstruction framework, SecretGen, to effectively recover private information. Compared with previous methods which can recover private data with the ground true prediction of the targeted recovery instance, SecretGen does not require such prior knowledge, making it more practical. We conduct extensive experiments on different datasets under diverse scenarios to compare SecretGen with other baselines and provide a systematic benchmark to better understand the impact of different auxiliary information and optimization operations. We show that without prior knowledge about true class prediction, SecretGen is able to recover private data with similar performance compared with the ones that leverage such prior knowledge. If the prior knowledge is given, SecretGen will significantly outperform baseline methods. We also propose several quantitative metrics to further quantify the privacy vulnerability of pre-trained models, which will help the model selection for privacy-sensitive applications. Our code is available at: https://github.com/AI-secure/SecretGen.