The Providentia project is led by professor Rasmus Pagh, hosted at DIKU, University of Copenhagen, and part of BARC. It is funded by a generous grant from Novo Nordisk Fonden.
The project runs from January 2022 to December 2026. The team will include (at least) 4 PhD students and 2 post-docs.
The aim of the project is to improve theory and practice of differentially private algorithms, with special emphasis on distributed settings. It revolves around four themes:
In all of these, algorithmic efficiency will be a core concern.
- Avoiding exchange of raw data
- Combining cryptographic primitives with differential privacy methods
- Privacy and robustness in machine learning
- Privacy-utility trade-offs, with focus on novel privacy models
Selected recent papers
- Differentially private sparse vectors with low error, optimal space, and fast access
Martin Aumüller, Christian Janos Lebeda, Rasmus Pagh.
Proceedings of Conference on Computer and Communications Security (CCS), 2021
- On the Power of Multiple Anonymous Messages
Badih Ghazi, Noah Golowich, Ravi Kumar, Rasmus Pagh, Ameya Velingker.
Proceedings of Eurocrypt, 2021
- Improved Utility Analysis of Private CountSketch
Rasmus Pagh, Mikkel Thorup.
Proceedings of Neural Information Processing Symposium (NeurIPS), 2022
- A Smooth Binary Mechanism for Efficient Private Continual Observation
Joel Daniel Andersson, Rasmus Pagh. Proceedings of Neural Information Processing Symposium (NeurIPS), 2023
Teaching and outreach
In ancient Rome the goddess Providentia personified the ability to foresee events and make suitable provision.
In this spirit, the Providentia project seeks to provide the forethought needed for scientists to make use of valuable sources of insight, even if data contains sensitive information.
The name is also a near-acronym of Privacy-driven Trust in Algorithms.