Validation of ML-UQ calibration statistics using simulated reference values: a sensitivity analysis

arXiv:2403.00423v1 Announce Type: cross
Abstract: Some popular Machine Learning Uncertainty Quantification (ML-UQ) calibration statistics do not have predefined reference values and are mostly used in comparative studies. In consequence, calibration is almost never validated and the diagnostic is left to the appreciation of the reader. Simulated reference values, based on synthetic calibrated datasets derived from actual uncertainties, have been proposed to palliate this problem. As the generative probability distribution for the simulation of synthetic errors is often not constrained, the …

abstract analysis arxiv cs.lg diagnostic machine machine learning physics.chem-ph popular quantification reader reference sensitivity statistics stat.ml studies type uncertainty validation values