AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Probably Approximately Correct Labels

Emmanuel Candes, Andrew Ilyas, Tijana Zrnic

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 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
add uncertainty calibration by group and bin (ablation)PAC labeling · Media bias (GPT-4o)Budget save (%): 13.68 -> 16.72Improved
route between GPT-4o and Claude with a PAC router instead of GPT-4o alonePAC labeling · Media biasBudget save (%): 13.79 -> 41.61Improved
route between GPT-4o and Claude with a PAC router instead of Claude alonePAC labeling · Media biasBudget save (%): 8.41 -> 41.61Improved
use cost-sensitive PAC router instead of GPT-4o alonePAC labeling · Media biasSave in cost (%): 188.66 -> 482.04Improved
use cost-sensitive PAC router instead of Claude alonePAC labeling · Media biasSave in cost (%): 131.36 -> 482.04Improved
use asymptotic instead of nonasymptotic confidence intervals (ablation)PAC labeling · Media biasBudget save (%): 13.79 -> 16.11Mixed
use asymptotic confidence intervals (ablation)PAC labeling · Stance on global warmingBudget save (%): 28.09 -> 32.15Mixed
use asymptotic confidence intervals (ablation)PAC labeling · MisinformationBudget save (%): 18.12 -> 21.41Mixed
use asymptotic confidence intervals (ablation)PAC labeling · ImageNetBudget save (%): 59.64 -> 62.82Mixed
use asymptotic confidence intervals (ablation)PAC labeling · ImageNet v2Budget save (%): 39.07 -> 39.20No clear effect
use PAC labeling instead of fixed threshold U>=0.1Naive threshold · Stance on global warmingError (%): 10.13 -> 4.57Improved
use PAC labeling instead of fixed threshold U>=0.1Naive threshold · MisinformationError (%): 7.07 -> 3.80Improved

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