arXiv stat.ML
· Papers
Linear-Core Surrogates: Smooth Loss Functions with Linear Rates for Classification and Structured Prediction
arXiv:2604.27742v2 Announce Type: replace-cross Abstract: A fundamental dichotomy in the theory of classification sets smoothness against statistical efficiency: smooth surrogate losses such as the logistic loss enable fast $O(1/T)$ optimization but yield slow square-root $H$-consistency bounds, while piecewise-linear