AI Paper Atlas

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

Online Control of the False Coverage Rate and False Sign Rate

Asaf Weinstein, Aaditya Ramdas

ICML 2020 · Learning theory and algorithms › Prediction and decision losses · approach: Probabilistic calibration · 4 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 LORD-CI marginal intervals instead of conditional intervalsOnline selective confidence intervals under a fixed-threshold selection rule · Fixed thresholdFCR: 0.1 -> 0.028Improved
Use LORD-CI marginal intervals instead of conditional intervalsOnline selective confidence intervals · Fixed thresholdSign-determining fraction: 0.502 -> 0.649Improved
Use LORD-CI rather than conditional intervalsSign-determining selection with symmetric intervals · Sign-determining LORD-CI (Symm)Sign-determining fraction: {0.533} -> 1Improved
Use LORD-CI rather than conditional intervalsSign-determining selection with MQC intervals · Sign-determining LORD-CI (MQC)Sign-determining fraction: 0.527 -> 1Improved
Use an MQC interval instead of a symmetric interval (ablation)Sign-determining LORD-CI · Simulation exampleSelections: 127 -> 144Improved
Use an MQC interval instead of a symmetric interval (ablation)Sign-determining LORD-CI · Simulation exampleFCP: 0.042No clear effect
Use sign-determining LORD-CI instead of conditional CIsSign-classification procedure · Sign-determining selection with symmetric intervalsSign-determining fraction: 51.2% of conditional CIs include both positive and negative valuesImproved
Discard conditional CIs that cross zero without readjusting (ablation)Conditional intervals after LORD++ selection · Conditional CIs following LORD++ selectionFCP: 14.5% -> 45.8%Hurt
Keep only conditional CIs that do not cross zero, then readjust them (ablation)Conditional intervals after LORD++ selection · Conditional CIs following LORD++ selectionFCP: 45.8% -> 0.125Improved
Use the MQC interval instead of the symmetric interval (ablation)Sign-determining LORD-CI · Sign-classification powerSelections: about 13% moreImproved

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