Concurrent Composition Theorems for all Standard Variants of Differential Privacy

Published on ● Video Link: https://www.youtube.com/watch?v=431JsJABRPk



Duration: 41:21
758 views
4


Wanrong Zhang (Harvard University)
https://simons.berkeley.edu/talks/concurrent-composition-theorems-all-standard-variants-differential-privacy
Societal Considerations and Applications

We study the concurrent composition properties of interactive differentially private mechanisms, whereby an adversary can arbitrarily interleave its queries to the different mechanisms. We prove that all composition theorems for non-interactive differentially private mechanisms extend to the concurrent composition of interactive differentially private mechanisms for all standard variants of differential privacy including $(\eps,\delta)$-DP with $\delta-0$, R\`enyi DP, and $f$-DP, thus answering the open question by \cite{vadhan2021concurrent}. For $f$-DP, which captures $(\eps,\delta)$-DP as a special case, we prove the concurrent composition theorems by showing that every interactive $f$-DP mechanism can be simulated by interactive post-processing of a non-interactive $f$-DP mechanism. For R\`enyi DP, we use a different approach by showing the optimal adversary against the concurrent composition can be decomposed as a product of the optimal adversaries against each interactive mechanism.




Other Videos By Simons Institute for the Theory of Computing


2022-11-10Supply-Side Equilibria in Recommender Systems
2022-11-10What Really Matters for Fairness in Machine Learning: Delayed Impact and Other Desiderata
2022-11-10Predictive Modeling in Healthcare – Special Considerations
2022-11-10Bringing Order to Chaos: Navigating the Disagreement Problem in Explainable ML
2022-11-09Pipeline Interventions
2022-11-09Algorithmic Challenges in Ensuring Fairness at the Time of Decision
2022-11-09Improving Refugee Resettlement
2022-11-09Learning to Predict Arbitrary Quantum Processes
2022-11-09A Kerfuffle: Differential Privacy and the 2020 Census
2022-11-08Chasing the Long Tail: What Neural Networks Memorize and Why
2022-11-08Concurrent Composition Theorems for all Standard Variants of Differential Privacy
2022-11-08Privacy Management: Achieving the Possimpible
2022-11-07Privacy-safe Measurement on the Web: Open Questions From the Privacy Sandbox
2022-10-29Transmission Neural Networks: From Virus Spread Models to Neural Networks
2022-10-29Spatial Spread of Dengue Virus: Appropriate Spatial Scales for Transmission
2022-10-28A Global Comparison of COVID-19 Variant Waves and Relationships with Clinical and...
2022-10-28Diversity and Inequality in Information Diffusion on Social Networks
2022-10-28Learning through the Grapevine and the Impact of the Breadth and Depth of Social Networks
2022-10-28Just a Few Seeds More: The Inflated Value of Network Data for Diffusion...
2022-10-27Bayesian Learning in Social Networks
2022-10-27Likelihood-based Inference for Stochastic Epidemic Models



Tags:
Simons Institute
theoretical computer science
UC Berkeley
Computer Science
Theory of Computation
Theory of Computing
Epidemics and Information Diffusion
Wanrong Zhang