AI Paper Atlas

Research map / Label-efficient and robust learning / Noisy labels

Label Noise Robustness of Conformal Prediction

Bat-Sheva Einbinder, Shai Feldman, Stephen Bates, Anastasios N. Angelopoulos, Asaf Gendler, Yaniv Romano

JMLR 2024 · Label-efficient and robust learning › Noisy labels · approach: Noise-aware loss and pairwise methods · 23 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
calibrate conformal prediction with noisy single-annotator labels instead of clean labelsVanilla conformal prediction with ResNet-18 · CIFAR-10Hcoverage of clean labels: 90% -> approximately 93%Improved
add uniform flip label noise (epsilon 0.05) to calibrationConformal prediction with oracle classifier · Synthetic multi-class classificationcoverage of clean labels: 90% -> around 94%Improved
use rare-to-most-frequent class noise instead of uniform flipConformal prediction with neural network · Synthetic multi-class classificationcoverage of clean labelsHurt
use APS score instead of HPS score under label noiseConformal prediction · Synthetic multi-class classificationrobustness of coverage to label noiseImproved
add dispersive or symmetric noise to responsesCQR conformal intervals · Synthetic regressioncoverage of clean responsesImproved
add biased or contractive noise to responsesCQR conformal intervals · Synthetic regressioncoverage of clean responsesHurt
fit predictive model on clean training data instead of noisy (ablation)CQR conformal intervals with biased noise · Synthetic regressioncoverage requirementImproved
calibrate with noisy labelsCQR and residual magnitude conformal intervals · AVA aesthetic regressioninterval width and coverageMixed
calibrate conformal risk control with noisy labels (fine and coarse)Conformal risk control with TResNet · Multi-label CIFAR-100NFNR on clean labelsNo clear effect
calibrate conformal risk control with noisy annotations (beta 0.1)Conformal risk control with PraNet · Polyp segmentationFNR on clean labelsNo clear effect
use distribution-free risk bounds on noisy calibrationConformal risk control, smooth miscoverage · Adversarial synthetic regressionsmooth miscoverage validityImproved
calibrate Rolling RC with noisy depth labelsRolling RC with LeReS · Depth estimation onlineimage miscoverage on clean labelsNo clear effect

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