{"id":"arxiv/2603.15232","url":"https://arxiv.org/abs/2603.15232","tags":["cs.LG","math.ST","stat.ML"],"score":0,"title":"Decomposing Probabilistic Scores: Reliability, Information Loss and Uncertainty","domain":"cs.LG","source":"arxiv","authors":["Arthur Charpentier","Agathe Fernandes Machado"],"reviews":0,"subject":"cs.LG","summary":"Calibration is a conditional property that depends on the information retained by a predictor. We develop decomposition identities for arbitrary proper losses that make this dependence explicit. At any information level $\\mathcal A$, the expected loss of an $\\mathcal A$-measurable predictor splits into a proper-regret (reliability) term and a conditional entropy (residual uncertainty) term. For nested levels $\\mathcal A\\subseteq\\mathcal B$, a chain decomposition quantifies the information gain from $\\mathcal A$ to $\\mathcal B$. Applied to classification with features $\\boldsymbol{X}$ and…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-03-16","reactions":{}}