Synth-MIA: A Testbed for Auditing Privacy Leakage in Tabular Data Synthesis
Joshua Ward, Xiaofeng Lin, Chi-Hua Wang, Guang Cheng · 2025
Read Synth-MIA: A Testbed for Auditing Privacy Leakage in Tabular Data Synthesis by Joshua Ward, Xiaofeng Lin, Chi-Hua Wang, Guang Cheng (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-18014.txt.
Description
Joshua Ward and 3 co-authors build a test kit to audit if fake data still leaks who was in the real set. Their 2025 tool, Synth-MIA, deals with tabular data, such as rows of patient or client facts. Fake sets are said to guard privacy, but plain likeness scores miss the real risk.
Attack tests that guess if a person was in the source set give a sharper read. Many such attacks exist, each in its own form, and none wins all the time. The kit runs a pack of attacks through one simple interface. It reports the worst leak found, so risk is not understated.
Catalog record
| CREATOR | Joshua Ward, Xiaofeng Lin, Chi-Hua Wang, Guang Cheng |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 70 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-18014