AI RESEARCH
Average Calibration Losses for Reliable Uncertainty in Medical Image Segmentation
arXiv CS.CV
•
ArXi:2506.03942v4 Announce Type: replace Deep neural networks for medical image segmentation are often overconfident, compromising both reliability and clinical utility. In this work, we propose differentiable formulations of marginal L1 Average Calibration Error (mL1-ACE) as an auxiliary loss that can be computed on a per-image basis. We compare both hard- and soft-binning approaches to directly improve pixel-wise calibration.