Center for Advanced Intelligence Project

[AIP AI Security and Privacy Team Seminar]

Locally Private Sampling: From Generative Models to Private Synthetic Data Generation

Speaker: Prof. Shahab Asoodeh (McMaster University, Canada)

Abstract

The problem of sampling under local differential privacy has recently gained attention due to its potential applications in generative models. However, a thorough understanding of the privacy-utility trade-off in this context is still lacking.

In this talk, I’ll discuss the minimax optimality of locally private sampling using f-divergences. I’ll demonstrate that this setup corresponds to the limiting case of distribution estimation under “user-level” local differential privacy, where each user has access to a large amount of data.

As the main result of the talk, I’ll present families of optimal sampling mechanisms for both discrete and continuous domains. Remarkably, these samplers are universally optimal across all f-divergences, distinguishing sampling from typical learning problems. If time permits, I’ll also discuss how non-private data can be integrated into private sampling problems. I’ll conclude the talk by highlighting several open questions in the area of private sampling.

Bio

Shahab Asoodeh is an Assistant Professor of computer science in the Department of Computing and Software at McMaster University and a Faculty Affiliate at Vector Institute. Before joining McMaster in August 2021, he was a visiting research scientist at Meta and a postdoctoral fellow at Harvard University. He received his Ph.D. in Applied Mathematics from Queen’s University, Canada. His research develops information-theoretic tools for trustworthy and reliable machine learning (including fair and privacy-preserving ML). He is currently a member of the diversity and inclusion committee of IEEE Information Theory Society.