Research Focus

My research focuses on how different types of uncertainty can be represented, quantified, and used in machine learning. In particular, I am interested in the reliability of deep neural networks, i.e., when can we trust their predictions, and how can models recognize and communicate when they do not know? A central goal of my work is to develop uncertainty-aware methods that address real-world problems and make machine learning systems more robust, transparent, and trustworthy. A related line of my research concerns conformal prediction, a framework for constructing statistically valid prediction sets under minimal assumptions. I am especially interested in online selective conformal inference, where data arrive sequentially and prediction sets or intervals are reported only at selected time points. More broadly, my research lies at the intersection of trustworthy artificial intelligence, statistical learning, and practical machine learning methodology.

Selected Publications