PhD student, Department of Computer Science, University of Toronto
1 paper at NeurIPS 2025
D-PDDM provably monitors model deterioration requiring no training data during deployment, and performs well in real-worlds datasets.
