AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Inference with non-differentiable surrogate loss in a general high-dimensional classification framework

Muxuan Liang, Yang Ning, Maureen A Smith, Ying-Qi Zhao

JMLR 2026 · Learning theory and algorithms › Prediction and decision losses · approach: Consistency of surrogate losses · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use proposed decorrelated-score inference instead of the ad-hoc refit methodLasso SVM decision rule · Diabetes HbA1c risk-factor classification (n=9101, p=120)average 95% CI length: 0.244 -> 0.122Improved
use proposed inference instead of the ad-hoc refit methodEstimated ITR · Diabetes optimal ITR (n=9101, p=120)average 95% CI length: 0.423 -> 0.122Improved
use estimated ITR instead of current clinical practiceType-II diabetes treatment · Diabetes optimal ITR (n=9101, p=120)success rate of HbA1c control: 0.860 -> 0.871Improved
apply decorrelation and sample splitting (proposed) instead of ad-hoc refit inferenceScenario I simulation · Simulation Scenario I (p 500-1600)Type-I errorImproved
use hinge loss instead of logistic lossProposed inference · Simulation Scenario I (p 500-1600)powerMixed
use proposed method instead of logistic-loss methodITR value function · Simulation Scenario II (p 500-1600)value functionImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy