Newest papers

Membership Inference Attack on Private Evolution

Daniil Filienko, Sikha Pentyala, Martine De Cock · 2026 · arXiv

DOI: 10.48550/arXiv.2610.30514 · open-access PDF

Korean

합성 데이터 생성(Synthetic Data Generation, SDG)은 개인 데이터의 안전한 공유를 더 광범위하게 가능하게 하는 적극적으로 연구된 방법입니다.Inference-only SDG methods, such as Private Evolution (PE), are appealing, as they can leverage black-box access to pre-trained foundation models to generate high-quality data without the need for model fine-tuning on the private data.PE is explicitly designed to produce synthetic data under Differential Privacy (DP).Now formal DP guarantees do not eliminate the need for empirical privacy audit of the practical susceptibility of released synthetic data to adversaria

Machine translation (Meta M2M100); the English original follows.

Abstract

Synthetic Data Generation (SDG) is an actively studied way to enable safe sharing of private data more broadly. Inference-only SDG methods, such as Private Evolution (PE), are particularly appealing, as they can leverage black-box access to pre-trained foundation models to generate high quality data without the need for model fine-tuning on the private data. PE is explicitly designed to produce synthetic data under Differential Privacy (DP). Yet formal DP guarantees do not eliminate the need for empirical privacy auditing of the practical susceptibility of released synthetic data to adversaria

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange