Spotlight Talks
- Aurélien Bellet | Inria
Author(s) | Tudor Cebere, Mathieu Even, Linus Bleistein, Aurélien Bellet
Privacy Auditing with Zero (0) Training Run - Aleksandar Nikolov | University of Toronto
Author(s) | Aleksandar Nikolov , Haohua Tang, Jonathan Ullman
Online Matrix Factorization and Online Private Query Release - Roodabeh Safavi | Institute of Science and Technology Austria (ISTA)
Author(s) | Monika Henzinger, Roodabeh Safavi, Salil Vadhan
Concurrent Composition for Differentially Private Continual Mechanisms - Teresa Anna Steiner | University of Southern Denmark
Author(s) | Giulia Bernardini , Philip Bille, Inge Li Gørtz, Teresa Anna Steiner
Differentially Private Substring and Document Counting with Near-Optimal Error - Thomas Steinke | Google DeepMind
Author(s) | Gunter F. Steinke, Thomas Steinke
Privately Estimating Black-Box Statistics - Marika Swanberg | Google Research
Author(s) | Marika Swanberg, Meenatchi Sundaram Muthu Selva Annamalai, Jamie Hayes, Borja Balle, Adam Smith
Beyond the Worst Case: Extending Differential Privacy Guarantees to Realistic Adversaries
Poster Presentations
- Joel Daniel Andersson | Institute of Science and Technology Austria (ISTA)
Author(s) | Joel Daniel Andersson, Palak Jain, Satchit Sivakumar
Improved Accuracy for Private Continual Cardinality Estimation
in Fully Dynamic Streams via Matrix Factorization - Bardiya Aryanfard | Institute of Science and Technology Austria (ISTA)
Author(s) | Bardiya Aryanfard, Monika Henzinger, David Saulpic, A.R. Sricharan
Improved Lower Bounds for Privacy under Continual Release - Fernando Beltran | The University of Auckland
Author(s) | Mina Khoshmehr, Fernando Beltran
Residual Privacy Budgeting with Weighted Scarcity Allocation for Online Query Answering - Simone Bombari | Institute of Science and Technology Austria (ISTA)
Author(s) | Simone Bombari, Jialei Luo, Inbar Seroussi, Marco Mondelli
High-Dimensional Private Linear Regression with Optimal Rates - Dennis Breutigam | University of Luebeck
Author(s) | Dennis Breutigam, Rüdiger Reischuk
Privacy Mechanisms against Adversaries with Limited Background Knowledge - Anamay Chaturvedi | Institute of Science and Technology Austria (ISTA)
Author(s) | Anamay Chaturvedi, Monika Henzinger Jalaj Upadhyay
Near-optimal generalized private testing - Edwige Cyffers | CNRS
- Limits of Personalizing Differential Privacy Budgets
Author(s) | Edwige Cyffers, Juba Ziani - Performative Privacy: When Differential Privacy Maximizes Utility
Author(s) | Uddalak Mukherjee, Edwige Cyffers
- Limits of Personalizing Differential Privacy Budgets
- Tamalika Mukherjee | Max Planck Institute for Security and Privacy
Author(s) | Abigail Gentle, Hendrik Fichtenberger,
Tamalika Mukherjee, Sayantan Sen
Private Graph Property Testing - Quentin Hillebrand | University of Copenhagen
Author(s) | Quentin Hillebrand, Jacob Imola, Rasmus Pagh, Sia Sejer
Unbiased Estimators from the Discrete Laplace Mechanism - Nikita Kalinin | Institute of Science and Technology Austria (ISTA)
Author(s) | Monika Henzinger, Nikita Kalinin ISTA, Jalaj Upadhyay
Normalized Square Root: Sharper Matrix Factorization Bounds for Differentially Private Continual Counting - Hannah Keller | Aarhus University
Author(s) | Jakob Burkhardt, Hannah Keller, Claudio Orlandi, Chris Schwiegelshohn
Distributed Differentially Private Data Analytics via Secure Sketching - Bogdan Kulynych | Biomedical Data Science Center, Lausanne University Hospital
Author(s) | Bogdan Kulynych, Antti Honkela
On Choosing the μ Parameter in Gaussian Differential Privacy - Christian Janos Lebeda | Inria, Idesp, Inserm, University of Montpellier
- Weighted Fourier Factorizations: Optimal Gaussian Noise for Differentially Private Marginal and Product Queries
Author(s) | Christian Janos Lebeda, Aleksandar Nikolov, Haohua Tang - Model Agnostic Differentially Private Causal Inference
Author(s) | Christian Janos Lebeda, Mathieu Even, Aurélien Bellet, Julie Josse
- Weighted Fourier Factorizations: Optimal Gaussian Noise for Differentially Private Marginal and Product Queries
- Johannes Liebenow | University of Luebeck
Author(s) | Johannes Liebenow, Thorsten Peinemann, Esfandiar Mohammadi
DP-Hype: Federated Differentially Private Hyperparameter Search - Andrew Lowy | CISPA Helmholtz Center for Information Security
Author(s) | Andrew Lowy, Daogao Liu
Differentially Private Bilevel Optimization: Efficient Algorithms with Near-Optimal Rates - Fredrik Meisingseth | Graz University of Technology
Author(s) | Fredrik Meisingseth, Christian Rechberger, Fabian Schmid
General Modularity Lemmata about Random Variable Commitment Schemes, and a Certified Laplace Mechanism - Clement Pierquin | Inria, Craft AI
Author(s) | Clément Pierquin, Aurélien Bellet, Marc Tommasi, Matthieu Boussard
Privacy Amplification Persists under Unlimited Data Release - Sia Sejer | BARC, University of Copenhagen
Author(s) | Rasmus Pagh, Sia Sejer
An Efficient Gaussian Mechanism under Continual Observation - Yara Schütt | University of Luebeck
Author(s) | Johannes Liebenow, Yara Schütt, Tanya Braun, Marcel Gehrke,
Florian Thaeter, Esfandiar Mohammadi
DPM : Clustering Sensitive Data through Separation - Tomer Shoham | The Hebrew University of Jerusalem
Author(s) | Tomer Shoham, Moshe Shenfeld, Noa Velner-Harris, Katrina Ligett
Differentially Private Nonparametric Confidence Intervals Under Minimal Distributional Assumptions - Sahel Torkamani | University of Edinburgh
Author(s) | Sahel Torkamani, Henry Gouk, Rik Sarkar
Privacy Requires Slicing: Differentially Private Magnitude via Projections - Long Tran | University of Helsinki
Author(s) | Long Tran, Antti Koskela, Ossi Räisä, Antti Honkela
f-Differential Privacy Filters: Validity and Approximate Solutions - Xuefeng Xu | University of Warwick
Author(s) | Xuefeng Xu, Graham Cormode
Federated Computation of ROC and PR Curves
Differential privacy has become the pre-eminent framework to measure and limit loss in privacy when statistics about sensitive data are computed and released. The theoretical study of differential privacy has extended far beyond this scope, establishing deep relationships with long studied areas of theoretical computer science, such as learning theory, robust algorithm design, adaptive data analysis and hypothesis testing. The goal of this workshop is to share and disseminate recent developments in the theory of differential privacy. We invite submissions of works dealing with the following topics:
- New differentially private mechanisms for wide variety of algorithmic problems superior to prior work
- Novel privacy accounting techniques and analyses
- Lower bounds/impossibility results related to differential privacy
- Relationships between differential privacy and other areas of TCS (for example formal methods)
Other closely allied topics may also be considered. Note that new mechanisms whose performance evaluation is purely empirical, without theoretical guarantees, are not within the scope of the workshop.
This workshop is non-archival and submission does not preclude submission at any future venues; works published prior are also welcome and encouraged. The reviews will be high-level, judging works based solely on novelty, interest, and relevance. Accepted submissions will be invited to present a poster in one of two poster sessions to be held at the workshop.
Submission Format
The submitted file should be a pdf, with at least 1 inch margins and a 10 point font, not including references. The file size should be at most 10 MB. Reviewers are only required to read the first 4 pages to assess the work, and may brief the rest of the material to judge the work at their discretion.The submission server will close at May 1st AOE, or until we reach capacity and author notifications will be made by June 1st. Please disclose whether an LLM was used for any part of the submission, and if so, in what manner.
Review Process
Each submission will be reviewed by members of the program committee and potential sub-reviewers. There is no expectation of extensive checks for correctness, just high-level sanity checks and most importantly an evaluation of interest and relevance to the workshop topics listed above. The submissions are not anonymous.