About
I am an Assistant Professor in the Department of Computer Science at Johns Hopkins University. I am a member of the Data Science and AI Institute (DSAI) and part of the Algorithms and Complexity group.
My research is on the theoretical foundations of trustworthy and reliable machine learning and statistics. I am particularly interested in differential privacy and its connections to robustness, generalization, and memorization in learning algorithms.
News
November 2026. If you've ever struggled to find the right statement of a fingerprinting lower bound for differential privacy, we have a solution to your problems! Check out our exposition on robust membership inference attacks, which aims to distill the key insights from important papers in the area. We were motivated to write it while working on our new paper on the impossiblity of dimension-free private mean estimation with unknown covariance.
May 2026. I am the Workshops, Tutorials, and Community Events Chair for COLT: please submit your proposals here!
January 2026. I am back at the Simons Institute for the Federated and Collaborative Learning program this Spring. Get in touch if you are in the Bay Area!
June 2025. Join our LeT-All community event at COLT featuring a fireside chat with Peter Bartlett and Vitaly Feldman, group activities, and mentorship roundtables.
June 2025. I'll give a keynote talk on directions in DP statistics at Theory and Practice of Differential Privacy 2025 .
Research
My research is on the theoretical foundations of trustworthy and reliable machine learning and statistics. I am particularly interested in differential privacy and its connections to robustness, generalization, and memorization in learning algorithms.
Selected Publications
-
Dimension-free Private Mean Estimation is Impossible Without the Covariance [arXiv]
Renaud Gaucher, Ephraim Linder, Adam Smith, Jonathan Ullman, Lydia Zakynthinou.
Manuscript.
-
Dimension-free Private Mean Estimation for Anisotropic Distributions [arXiv]
Yuval Dagan, Xuelin Yang, Michael I. Jordan, Lydia Zakynthinou, and Nikita Zhivotovskiy.
NeurIPS 2024.
-
From Robustness to Privacy and Back [arXiv]
Hilal Asi, Jonathan Ullman, and Lydia Zakynthinou.
ICML 2023.
-
Covariance-Aware Private Mean Estimation Without Private Covariance Estimation [arXiv]
Gavin Brown, Marco Gaboardi, Adam Smith, Jonathan Ullman, and Lydia Zakynthinou.
NeurIPS 2021. Spotlight presentation.
-
Private Identity Testing for High-Dimensional Distributions [arXiv]
Clement L. Canonne, Gautam Kamath, Audra McMillan, Jonathan Ullman, and Lydia Zakynthinou.
NeurIPS 2020. Spotlight presentation.
-
Reasoning About Generalization via Conditional Mutual Information [arXiv]
Thomas Steinke and Lydia Zakynthinou.
COLT 2020.
Mentoring
I hold weekly office hours as part of an initiative by the Learning Theory Alliance. You're welcome to book a time to chat with me here.
Current Group
- Michael Xie (PhD)
- Chengyuan Deng: DSAI Postdoc fellow, co-hosted by Mike Dinitz and Jess Sorrell
Alumni
- Renaud Gaucher: PhD intern at UC Berkeley
Teaching
Courses
- EN.601.438/638: Theory of Differential Privacy, Fall 2026.
- EN.601.488/688: Foundations of Responsible AI, Spring 2027.
Tutorials
- Tutorial on differential privacy at the Australasian Summer School on Recent Trends in Algorithms.
- Advanced session on differential privacy for the Berkeley Math Circle.
Service
- Workshop committee member for the Learning Theory Alliance.
- Program and reviewing committee member for IEEE S&P, ICML, NeurIPS, TPDP, AAAI, FAccT, and COLT.
CV
Short bio
Lydia Zakynthinou is an Assistant Professor in the Department of Computer Science at Johns Hopkins University and a member of its Data Science and AI Institute (DSAI). She received an Electrical and Computer Engineering diploma and a MSc in Logic, Algorithms, and Theory of Computation from the National Technical University of Athens, and her PhD from Northeastern University, advised by Jonathan Ullman and Huy Lê Nguyễn. She was a FODSI postdoctoral research fellow at the Simons Institute for the Theory of Computing at UC Berkeley, hosted by Michael I. Jordan. Her research is on theoretical foundations of trustworthy machine learning and statistics, data privacy, and robustness.