{"id":"arxiv/2606.19147","url":"https://arxiv.org/abs/2606.19147","tags":["stat.ML","cs.LG","math.ST"],"score":0,"title":"Cross-Calibrated Confidence Fields for Local Risk Updates","domain":"stat.ML","source":"arxiv","authors":["Mingzhi Song"],"reviews":0,"subject":"stat.ML","summary":"How can training data be used to compare local updates to the current model, choose an update, and retain valid bounds for the selected update's population-risk change? We construct lower and upper confidence fields that jointly cover the population-risk change of every update in a possibly continuous local update space. The fields can therefore be used both to compare the updates and to select an update; a negative upper endpoint certifies improvement over the reference model. For linear risk changes in possibly infinite-dimensional feature spaces, cross-calibration uses the discrepancy…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-06-17","reactions":{}}