Limits of Private Learning with Access to Public Data

Abstract

We consider learning problems where the training set consists of two types of examples: private and public. The goal is to design a learning algorithm that satisfies differential privacy only with respect to the private examples. This setting interpolates between private learning (where all examples are private) and classical learning (where all examples are public). We study the limits of learning in this setting in terms of private and public sample complexities. We show that any hypothesis class of VC-dimension $d$ can be agnostically learned up to an excess error of $\alpha$ using only (roughly) $d/\alpha$ public examples and $d/{\alpha }^2$ private labeled examples. This result holds even when the public examples are unlabeled. This gives a quadratic improvement over the standard $d/{\alpha }^2$ upper bound on the public sample complexity (where private examples can be ignored altogether if the public examples are labeled). Furthermore, we give a nearly matching lower bound, which we prove via a generic reduction from this setting to the one of private learning without public data.

Cited by in corpus (10)

• Differentially Private Fine-tuning of Language Models
• Do Not Let Privacy Overbill Utility: Gradient Embedding Perturbation for Private Learning
• Leveraging Public Data for Practical Private Query Release
• Reasoning About Generalization via Conditional Mutual Information
• Bypassing the Ambient Dimension: Private SGD with Gradient Subspace Identification
• Learning from Mixtures of Private and Public Populations
• PAC learning with stable and private predictions
• Public Data-Assisted Mirror Descent for Private Model Training
• Revisiting Model-Agnostic Private Learning: Faster Rates and Active Learning
• Realizable Learning is All You Need