Our work Optimal Zero-shot Regret Minimization for Selective Classification with Out-of-Distribution Detection was accepted at UAI 2025, The 41st Conference on Uncertainty in Artificial Intelligence.
A huge shout-out to my great colleague Eduardo for his fantastic work!
Marco Romanelli
Research Associate
My research interests include applications of Information Theory notions to Privacy and Security, Safety in AI, Machine Learning and Information Leakage Measurement.