Fine-Tuning with Differential Privacy Necessitates an Additional Hyperparameter Search
Cattan, Yannis, Choquette-Choo, Christopher A., Papernot, Nicolas, Thakurta, Abhradeep
–arXiv.org Artificial Intelligence
Models need to be trained with privacy-preserving learning algorithms to prevent leakage of possibly sensitive information contained in their training data. However, canonical algorithms like differentially private stochastic gradient descent (DP-SGD) do not benefit from model scale in the same way as non-private learning. This manifests itself in the form of unappealing tradeoffs between privacy and utility (accuracy) when using DP-SGD on complex tasks. To remediate this tension, a paradigm is emerging: fine-tuning with differential privacy from a model pretrained on public (i.e., non-sensitive) training data. In this work, we identify an oversight of existing approaches for differentially private fine tuning. They do not tailor the fine-tuning approach to the specifics of learning with privacy. Our main result is to show how carefully selecting the layers being fine-tuned in the pretrained neural network allows us to establish new state-of-the-art tradeoffs between privacy and accuracy. For instance, we achieve 77.9% accuracy for $(\varepsilon, \delta)=(2, 10^{-5})$ on CIFAR-100 for a model pretrained on ImageNet. Our work calls for additional hyperparameter search to configure the differentially private fine-tuning procedure itself.
arXiv.org Artificial Intelligence
Oct-5-2022
- Country:
- North America > United States
- California > San Diego County > San Diego (0.04)
- Asia > Japan
- Kyūshū & Okinawa > Okinawa (0.04)
- North America > United States
- Genre:
- Research Report (0.40)
- Industry:
- Information Technology > Security & Privacy (1.00)
- Technology: