Better Membership Inference Privacy Measurement through Discrepancy
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914809829654528 |
|---|---|
| author | Wu, Ruihan Huang, Pengrun Chaudhuri, Kamalika |
| author_facet | Wu, Ruihan Huang, Pengrun Chaudhuri, Kamalika |
| contents | Membership Inference Attacks have emerged as a dominant method for empirically measuring privacy leakage from machine learning models. Here, privacy is measured by the {\em{advantage}} or gap between a score or a function computed on the training and the test data. A major barrier to the practical deployment of these attacks is that they do not scale to large well-generalized models -- either the advantage is relatively low, or the attack involves training multiple models which is highly compute-intensive. In this work, inspired by discrepancy theory, we propose a new empirical privacy metric that is an upper bound on the advantage of a family of membership inference attacks. We show that this metric does not involve training multiple models, can be applied to large Imagenet classification models in-the-wild, and has higher advantage than existing metrics on models trained with more recent and sophisticated training recipes. Motivated by our empirical results, we also propose new membership inference attacks tailored to these training losses. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_15140 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Better Membership Inference Privacy Measurement through Discrepancy Wu, Ruihan Huang, Pengrun Chaudhuri, Kamalika Machine Learning Membership Inference Attacks have emerged as a dominant method for empirically measuring privacy leakage from machine learning models. Here, privacy is measured by the {\em{advantage}} or gap between a score or a function computed on the training and the test data. A major barrier to the practical deployment of these attacks is that they do not scale to large well-generalized models -- either the advantage is relatively low, or the attack involves training multiple models which is highly compute-intensive. In this work, inspired by discrepancy theory, we propose a new empirical privacy metric that is an upper bound on the advantage of a family of membership inference attacks. We show that this metric does not involve training multiple models, can be applied to large Imagenet classification models in-the-wild, and has higher advantage than existing metrics on models trained with more recent and sophisticated training recipes. Motivated by our empirical results, we also propose new membership inference attacks tailored to these training losses. |
| title | Better Membership Inference Privacy Measurement through Discrepancy |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2405.15140 |