Fabrizio Dabbene

Directional Conformal Uncertainty Quantification from Learned Model Discrepancy

Cesare Donati, Fabrizio Dabbene, Martina Mammarella

We propose a conformal prediction framework for quantifying the error of physics-based predictors used in control, where simple models are preferred for synthesis, certification, and real-time use. Because these models are selected for compatibility with the intended application…Read more