The Discrete Gaussian for Differential Privacy
Clément L. Canonne, Gautam Kamath, et al.
NeurIPS 2020
The article describes a practical method for detecting outlier database connections in real-time. Outlier connections are detected with a specified level of confidence. The method is based on generalized security rules and a simple but effective real-time machine learning mechanism. The described method is non-intrusive to the database and does not depend on the type of database. The method is used to proactively control access even before database connection is established, minimize false positives, and maintain the required response speed to detected database connection outliers. The capabilities of the system are demonstrated with several examples of outliers in real-world scenarios.
Clément L. Canonne, Gautam Kamath, et al.
NeurIPS 2020
Aladin Djuhera, Swanand Ravindra Kadhe, et al.
ICLR 2025
Chulin Xie, Keli Huang, et al.
ICLR 2020
Dan Williams, Milo Craun, et al.
SOSP 2025