Poster
PANORAMIA: Efficient Privacy Auditing of Machine Learning Models without Retraining
Mishaal Kazmi · Hadrien Lautraite · Alireza Akbari · Qiaoyue Tang · Mauricio Soroco · Tao Wang · Sébastien Gambs · Mathias Lécuyer
West Ballroom A-D #6009
Abstract:
We introduce PANORAMIA, a privacy leakage measurement scheme for Machine Learning (ML) models that relies on membership inference attacks using generated data as “non-members”. PANORAMIA does not modify the training data or training process, and only requires access to a subset of the training data. We evaluate PANORAMIA on ML models for image and tabular data classification, and on large-scale language models.
Live content is unavailable. Log in and register to view live content