Research Papers

I am committed to making all of my papers freely available as quickly as possible by hosting them on arXiv, and these preprints should be considered the authoritative version of the paper. When possible I will also link to any associated code, data, talks, or expository writing available. These resources are indicated with paper talk code misc tags.

I prefer not to order authors by contribution, but sometimes the norms of the community and the needs of students demand it. Author ordering is alphabetical except where otherwise noted.

Manuscripts

Efficient and Optimal Learning of Discrete Distributions with Person-Level Privacy

Gautam Kamath, Mahbod Majid, Ankur Moitra, Argyris Mouzakis, and Jonathan Ullman

Preprint

The Sample Complexity of Membership Inference and Privacy Auditing arXiv

Mahdi Haghifam, Adam Smith, and Jonathan Ullman

Preprint

Instance-Optimal Differentially Private Estimation arXiv

Audra McMillan, Adam Smith, and Jonathan Ullman

Preprint

Conference and other Primary Publications

Like most computer scientists, my work is primarily published in competitive conferences, which are typically as selective as top journals. This list includes all primary publications of my work in regardless of venue.

2026

Online Matrix Factorization, Online Private Query Release, and Online Discrepancy Minimization

Sasho Nikolov, Haohua Tang, and Jonathan Ullman

STOC ’26 ACM Symposium on Theory of Computing

Black-Box Privacy Attacks on Shared Representations in Multitask Learning arXiv

John Abascal, Nicolás Berrios, Alina Oprea, Jonathan Ullman, Adam Smith, and Matthew Jagielski (contribution order)

ICLR ’26 International Conference on Learning Representations

Lower Bounds for Public-Private Learning Under Distribution Shift arXiv

Amrith Setlur, Pratiksha Thaker, and Jonathan Ullman

AISTATS ’26 International Conference on Artificial Intelligence and Statistics

2025

A Bias-Variance-Privacy Trilemma for Statistical Estimation arXiv

Gautam Kamath, Argyris Mouzakis, Matthew Regehr, Vikrant Singhal, Thomas Steinke, and Jonathan Ullman

Journal of the American Statistical Association ’25

Privacy in Metalearning and Multitask Learning: Modeling and Separations arXiv

Maryam Aliakbarpour, Konstantina Bairaktari, Adam Smith, Marika Swanberg, and Jonathan Ullman

AISTATS ’25 International Conference on Artificial Intelligence and Statistics

Private Mean Estimation with Person-Level Differential Privacy arXiv

Sushant Agarwal, Gautam Kamath, Mahbod Majid, Argyris Mouzakis, Rose Silver, and Jonathan Ullman

SODA ’25 ACM-SIAM Symposium on Discrete Algorithms

2024

Private Geometric Median arXiv

Mahdi Haghifam, Thomas Steinke, and Jonathan Ullman

NeurIPS ’24 Conference on Neural and Information Processing Systems

Metalearning with Very Few Samples Per Task arXiv

Maryam Aliakbarpour, Konstantina Bairaktari, Gavin Brown, Adam Smith, and Jonathan Ullman

COLT ’24 Conference on Learning Theory

Smooth Lower Bounds for Differentially Private Algorithms via Padding-and-Permuting Fingerprinting Codes arXiv

Naty Peter, Eliad Tsfadia, and Jonathan Ullman

COLT ’24 Conference on Learning Theory

How to Make the Gradients Small Privately: Improved Rates for Differentially Private Non-Convex Optimization arXiv

Andrew Lowy, Jonathan Ullman, and Stephen Wright

ICML ’24 International Conference on Machine Learning

Program Analysis for Adaptive Data Analysis

Jiawen Liu, Weihao Qu, Marco Gaboardi, Deepak Garg, and Jonathan Ullman (contribution order)

PLDI ’24 Conference on Programming Languages Design and Implementation

TMI! Finetuned Models Leak Private Information from their Pretraining Data arXiv

John Abascal, Stanley Wu, Alina Oprea, and Jonathan Ullman (contribution order)

PETS ’24 Privacy Enhancing Technologies Symposium

Chameleon: Increasing Label-Only Membership Leakage with Adaptive Poisoning arXiv Video GitHub

Harsh Chaudhuri, Giorgio Severi, Alina Oprea, and Jonathan Ullman (contribution order)

ICLR ’24 International Conference on Learning Representations

Differentially Private Medians and Interior Points for Non-Pathological Data arXiv

Maryam Aliakbarpour, Rose Silver, Thomas Steinke, and Jonathan Ullman

ITCS ’24 Innovations in Theoretical Computer Science

2023

Fair and Useful Cohort Selection arXiv

Konstantina Bairaktari, Paul Langton, Huy Le Nguyen, Niklas Smedemark-Margulies, and Jonathan Ullman

TMLR ’23 Transactions on Machine Learning Research

Visual Utility Evaluation of Differentially Private Scatterplots IEEE Xplore

Liudas Panavas, Tarik Crnovrsanin, Jane L. Adams, Ali Sarvghad, Melanie Tory, Jonathan Ullman, and Cody Dunne (contribution order)

IEEE TVCG ’23 IEEE Transactions on Visualization and Computer Graphics

Multitask Learning via Shared Features: Algorithms and Hardness arXiv

Konstantina Bairaktari, Guy Blanc, Li-Yang Tan, Jonathan Ullman, and Lydia Zakynthinou

COLT ’23 Conference on Learning Theory

From Robustness to Privacy and Back arXiv

Hilal Asi, Jonathan Ullman, and Lydia Zakynthinou

ICML ’23 International Conference on Machine Learning

How to Combine Membership-Inference Attacks on Multiple Updated Models arXiv

Matthew Jagielski*, Stanley Wu*, Alina Oprea, Jonathan Ullman, and Roxana Geambasu (contribution order)

PETS ’23 Privacy Enhancing Technologies Symposium

SNAP: Efficient Extraction of Private Properties with Poisoning arXiv 15m Video GitHub

Harsh Chaudhuri, John Abascal, Alina Oprea, Matthew Jagielski, Florian Tramèr, and Jonathan Ullman (contribution order)

IEEE S&P ’23 IEEE Symposium on Security and Privacy

2022

A Private and Computationally Efficient Estimator for Unbounded Gaussians arXiv

Gautam Kamath, Argyris Mouzakis, Vikrant Singhal, Thomas Steinke, and Jonathan Ullman

COLT ’22 Conference on Learning Theory

2021

Covariance-Aware Private Mean Estimation Without Private Covariance Estimation arXiv

Gavin Brown, Marco Gaboardi, Adam Smith, Jonathan Ullman, and Lydia Zakynthinou

NeurIPS ’21 Conference on Neural and Information Processing Systems

Selected as a Spotlight Presentation

Leveraging Public Data for Practical Private Query Release arXiv 60m Video

Terrance Liu, Thomas Steinke, Jonathan Ullman, Giuseppi Vietri, and Zhiwei Steven Wu

ICML ’21 International Conference on Machine Learning

The Limits of Pan Privacy and Shuffle Privacy for Learning and Estimation arXiv 60m Video

Albert Cheu and Jonathan Ullman

STOC ’21 ACM Symposium on Theory of Computing

Manipulation Attacks in Local Differential Privacy arXiv 1m Video

Albert Cheu, Adam Smith, and Jonathan Ullman

IEEE S&P ’21 IEEE Symposium on Security and Privacy

2020

Auditing Differentially Private Machine Learning: How Private is Private SGD? arXiv 15m Video GitHub

Matthew Jagielski, Jonathan Ullman, and Alina Oprea (contribution order)

NeurIPS ’20 Conference on Neural and Information Processing Systems

CoinPress: Practical Private Mean and Covariance Estimation arXiv GitHub

Sourav Biswas, Yihe Dong, Gautam Kamath, and Jonathan Ullman

NeurIPS ’20 Conference on Neural and Information Processing Systems

Private Identity Testing for High-Dimensional Distributions arXiv

Clément Canonne, Gautam Kamath, Audra McMillan, Jonathan Ullman, and Lydia Zakynthinou

NeurIPS ’20 Conference on Neural and Information Processing Systems

Selected as a Spotlight Presentation

Private Query Release Assisted by Public Data arXiv 15m Video

Raef Bassily, Albert Cheu, Shay Moran, Aleksandar Nikolov, Jonathan Ullman, and Zhiwei Steven Wu

ICML ’20 International Conference on Machine Learning

Private Mean Estimation of Heavy-Tailed Distributions arXiv

Gautam Kamath, Vikrant Singhal, and Jonathan Ullman

COLT ’20 Conference on Learning Theory

The Power of Factorization Mechanisms in Local and Central Differential Privacy arXiv

Alexander Edmonds, Aleksandar Nikolov, and Jonathan Ullman

STOC ’20 ACM Symposium on Theory of Computing

Efficient Private Algorithms for Learning Large-Margin Halfspaces arXiv

Huy Lê Nguyễn, Jonathan Ullman, and Lydia Zakynthinou

ALT ’20 Conference on Algorithmic Learning Theory

2019

Differentially Private Algorithms for Learning Mixtures of Gaussians arXiv

Gautam Kamath, Or Sheffet, Vikrant Singhal, and Jonathan Ullman

NeurIPS ’19 Conference on Neural and Information Processing Systems

Efficiently Estimating Erdős-Rényi Graphs with Differential Privacy arXiv Poster

Adam Sealfon and Jonathan Ullman

NeurIPS ’19 Conference on Neural and Information Processing Systems

Securely Sampling Biased Coins with Applications to Differential Privacy ePrint

Jeffrey Champion, abhi shelat, and Jonathan Ullman

CCS ’19 ACM Conference on Computer Security

Differentially Private Fair Learning arXiv

Matthew Jagielski, Michael Kearns, Jieming Mao, Alina Oprea, Aaron Roth, Saeed Sharifi-Malvajerdi, and Jonathan Ullman

ICML ’19 International Conference on Machine Learning

Privately Learning High-Dimensional Distributions arXiv Talk Video

Gautam Kamath, Jerry Li, Vikrant Singhal, and Jonathan Ullman

COLT ’19 Conference on Learning Theory

The Structure of Optimal Private Tests for Simple Hypotheses arXiv Talk Video

Clément Canonne, Gautam Kamath, Audra McMillan, Adam Smith, and Jonathan Ullman

STOC ’19 ACM Symposium on Theory of Computing

Distributed Differential Privacy via Shuffling arXiv

Albert Cheu, Adam Smith, Jonathan Ullman, David Zeber, and Maxim Zhilyaev

EUROCRYPT ’19 IACR International Conference on Theory and Application of Cryptographic Techniques

2018

The Fienberg Problem: How to Allow Human Interactive Data Analysis in the Age of Differential Privacy

Cynthia Dwork and Jonathan Ullman

Journal of Privacy and Confidentiality '18

Computing Marginals Using MapReduce arXiv

Foto Afrati, Shantanu Sharma, Jeffrey Ullman, and Jonathan Ullman

Journal of Computer and Systems Science ’18

Local Differential Privacy for Evolving Data arXiv

Matthew Joseph, Aaron Roth, Jonathan Ullman, and Bo Waggoner

NeurIPS ’18 Conference on Neural and Information Processing Systems

Selected as a Spotlight Presentation

The Limits of Post-Selection Generalization arXiv

Kobbi Nissim, Adam Smith, Thomas Steinke, Uri Stemmer, and Jonathan Ullman

NeurIPS ’18 Conference on Neural and Information Processing Systems

Hardness of Non-Interactive Differential Privacy from One-Way Functions ePrint

Lucas Kowalczyk, Tal Malkin, Jonathan Ullman, and Daniel Wichs

CRYPTO ’18 IACR International Cryptology Conference

Skyline Identification in Multi-Armed Bandits arXiv

Albert Cheu, Ravi Sundaram, and Jonathan Ullman

ISIT ’18 IEEE International Symposium on Information Theory

2017

An Antifolk Theorem for Large Repeated Games arXiv

Mallesh Pai, Aaron Roth, and Jonathan Ullman

ACM Transactions on Economics and Computation ’17

Between Pure and Approximate Differential Privacy arXiv

Thomas Steinke and Jonathan Ullman

Journal of Privacy and Confidentiality ’17

Tight Bounds for Differentially Private Selection arXiv

Thomas Steinke and Jonathan Ullman

FOCS ’17 IEEE Symposium on Foundations of Computer Science

Fractional Set Cover in the Streaming Model arXiv

Piotr Indyk, Sepideh Mahabadi, Ronitt Rubinfeld, Jonathan Ullman, Ali Vakilian, and Anak Yodpinyanee

APPROX ’17 International Workshop on Approximation Algorithms for Combinatorial Optimization Problems

The Price of Selection in Differential Privacy arXiv

Mitali Bafna and Jonathan Ullman

COLT ’17 Conference on Computational Learning Theory

Multidimensional Dynamic Pricing for Welfare Maximization arXiv

Aaron Roth, Aleksandrs Slivkins, Jonathan Ullman, and Zhiwei Steven Wu

EC ’17 ACM Conference on Economics and Computation

Invited to a special issue of Transactions on Economics and Computation for EC ’17

Make Up Your Mind: The Price of Online Queries in Differential Privacy arXiv

Mark Bun, Thomas Steinke, Jonathan Ullman

SODA ’17 ACM-SIAM Symposium on Discrete Algorithms

2016

Privacy Odometers and Filters: Pay-as-you-go Composition arXiv

Ryan Rogers, Aaron Roth, Jonathan Ullman, and Salil Vadhan

NeurIPS ’16 Conference on Neural and Information Processing Systems

Strong Hardness of Privacy from Weak Traitor Tracing arXiv

Lucas Kowalczyk, Tal Malkin, Jonathan Ullman, and Mark Zhandry

TCC ’16B IACR Theory of Cryptography Conference

Space Lower Bounds for Itemset Frequency Sketches arXiv

Edo Liberty, Michael Mitzenmacher, Justin Thaler, and Jonathan Ullman

PODS ’16 ACM Symposium on Principles of Database Systems

Algorithmic Stability for Adaptive Data Analysis arXiv

Raef Bassily, Kobbi Nissim, Uri Stemmer, Adam Smith, Thomas Steinke, and Jonathan Ullman

STOC ’16 ACM Symposium on Theory of Computing

Invited to a special issue of SIAM Journal on Computing for STOC ’16

Watch and Learn: Optimizing from Revealed Preferences Feedback arXiv SIGEcom Exchanges

Aaron Roth, Jonathan Ullman, and Zhiwei Steven Wu

STOC ’16 ACM Symposium on Theory of Computing

2015

When Can Limited Randomness Be Used in Repeated Games? arXiv

Pavel Hubác̆ek, Moni Naor, Jonathan Ullman

SAGT ’15 International Symposium on Algorithmic Game Theory

Invited to a special issue of Theory of Computing for SAGT ’15

Robust Traceability from Trace Amounts PDF

Cynthia Dwork, Adam Smith, Thomas Steinke, Jonathan Ullman, and Salil Vadhan

FOCS ’15 IEEE Symposium on Foundations of Computer Science

Interactive Fingerprinting Codes and the Hardness of Preventing False Discovery arXiv

Thomas Steinke and Jonathan Ullman

COLT ’15 Conference on Computational Learning Theory

Inducing Approximately Optimal Flow Using Truthful Mediators arXiv

Ryan Rogers, Aaron Roth, Jonathan Ullman, Zhiwei Steven Wu

EC ’15 ACM Conference on Economics and Computation

Private Multiplicative Weights Beyond Linear Queries arXiv

Jonathan Ullman

PODS ’15 ACM Symposium on Principles of Database Systems

2014

Preventing False Discovery in Interactive Data Analysis is Hard arXiv

Moritz Hardt and Jonathan Ullman

FOCS ’14 IEEE Symposium on Foundations of Computer Science

Privately Solving Linear Programs arXiv

Justin Hsu, Aaron Roth, Tim Roughgarden, Jonathan Ullman

ICALP ’14 International Colloquium on Automata, Languages, and Programming Track A

Fingerprinting Codes and the Price of Approximate Differential Privacy arXiv

Mark Bun, Jonathan Ullman, and Salil Vadhan

STOC ’14 ACM Symposium on Theory of Computing

Invited to a special issue of SIAM Journal of Computing for STOC ’14

Faster Private Release of Marginals on Small Databases arXiv

Karthekeyan Chandrasekaran, Justin Thaler, Jonathan Ullman, and Andrew Wan

ITCS ’14 Innovations in Theoretical Computer Science

Robust Mediators in Large Games arXiv

Michael Kearns, Mallesh Pai, Ryan Rogers, Aaron Roth, and Jonathan Ullman

ITCS ’14 Innovations in Theoretical Computer Science

2013

Differential Privacy for the Analyst via Private Equilibrium Computation arXiv

Justin Hsu, Aaron Roth, and Jonathan Ullman

STOC ’13 ACM Symposium on Thoery of Computing

Answering n2+o(1) Counting Queries with Differential Privacy is Hard arXiv

Jonathan Ullman

STOC ’13 ACM Symposium on Thoery of Computing

Invited to a special issue of SIAM Journal of Computing for STOC ’13

2012

Faster Algorithms for Privately Releasing Marginals arXiv

Justin Thaler, Jonathan Ullman, and Salil Vadhan

ICALP ’12 International Colloquium on Automata, Languages, and Programming Track A

Iterative Constructions and Private Data Release arXiv

Anupam Gupta, Aaron Roth, and Jonathan Ullman

TCC ’12 IACR Theory of Cryptography Conference

Privately Releasing Conjunctions and the Statistical Query Barrier arXiv

Anupam Gupta, Moritz Hardt, Aaron Roth, and Jonathan Ullman

STOC ’12 ACM Symposium on Theory of Computing

2011

PCPs and the Hardness of Generating Private Synthetic Data ECCC

Jonathan Ullman and Salil Vadhan

TCC ’11 IACR Theory of Cryptography Conference

Invited to the Journal of Cryptology

2010

Course Allocation by Proxy Auction

Scott Kominers, Mike Ruberry, Jonathan Ullman

WINE ’10 Workshop on Internet and Network Economics

The Price of Privately Releasing Contingency Tables and the Spectra of Random Matrices with Correlated Rows

Shiva Kasiviswanathan, Mark Rudelson, Adam Smith, Jonathan Ullman

STOC ’10 ACM Symposium on Theory of Computing

Secondary Publications

Some of my work appears in journals as a secondary form of publication, after initially appearing in a computer science conference. This list includes all such publications.

Manipulation Attacks in Local Differential Privacy arXiv 1m Video

Albert Cheu, Adam Smith, and Jonathan Ullman

Journal of Privacy and Confidentiality ’21

Efficiently Estimating Erdős–Rényi Graphs with Differential Privacy arXiv Poster

Adam Sealfon and Jonathan Ullman

Journal of Privacy and Confidentiality ’21

PCPs and the Hardness of Generating Private Synthetic Data ECCC

Jonathan Ullman and Salil Vadhan

JoC ’20 Journal of Cryptology

Invited submission from TCC ’11

Multidimensional Dynamic Pricing for Welfare Maximization

Aaron Roth, Aleksandrs Slivkins, Jonathan Ullman, and Zhiwei Steven Wu

TEAC ’20 ACM Transactions on Economics and Computation

Special issue for invited papers from EC ’17

Fingerprinting Codes and the Price of Approximate Differential Privacy arXiv

Mark Bun, Jonathan Ullman, and Salil Vadhan

SICOMP ’18 SIAM Journal on Computing

Special issue for invited papers from STOC ’14

When Can Limited Randomness Be Used in Repeated Games? arXiv

Pavel Hubáček, Moni Naor, and Jonathan Ullman

ToCS ’16 Theory of Computing Systems

Special issue for invited papers from SAGT ’15

Answering n²+o(1) Counting Queries with Differential Privacy is Hard arXiv

Jonathan Ullman

SICOMP ’16 SIAM Journal on Computing

Special issue for invited papers from STOC ’13

Privately Releasing Conjunctions and the Statistical Query Barrier arXiv

Anupam Gupta, Moritz Hardt, Aaron Roth, and Jonathan Ullman

SICOMP ’13 SIAM Journal on Computing

Surveys and Other Writing

DifferentialPrivacy.org Website

Blog

A Primer on Private Statistics Blog Post PDF

Gautam Kamath and Jonathan Ullman

Subgaussian Concentration via Stability Arguments arXiv

Thomas Steinke and Jonathan Ullman

Technical Perspective: Building a Safety Net for Data Reuse CACM

Jonathan Ullman

Communications of the ACM’17

PSI (Ψ): A Private data Sharing Interface arXiv

Marco Gaboardi, James Honaker, Gary King, Jack Murtagh, Kobbi Nissim, Jonathan Ullman, and Salil Vadhan

Exposed! A Survey of Attacks on Private Data ARSA

Cynthia Dwork, Adam Smith, Thomas Steinke, and Jonathan Ullman

Annual Review of Statistics and its Applications '17