AI Paper Atlas

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

Stable Localized Conformal Prediction via Transduction

Yinjie Min, Liuhua Peng, Changliang Zou

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Confidence calibration objectives · 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 StCP transduction with source-task data and unlabeled target dataGLCP-type conformal · CRIME real dataset-size std: 0.75 -> 0.50Improved
use StCPGLCP-type conformal · STAR real dataset-size std: 8.78 -> 5.25Improved
use StCPGLCP-type conformal · BIO real dataset-size std: 0.53 -> 0.37Improved
use StCPCQR-type conformal · STAR real dataset-size std: 6.45 -> 5.90Improved
use StCPCQR-type conformal · DERMA real dataset-size std: 0.15 -> 0.14Improved
use data-driven lambda selection (StCP-sel) instead of best-grid lambda (ablation)StCP GLCP-type · CRIME real dataset-size std: 0.50 -> 0.58Hurt
estimate S|X from source data with PPI-style debiasing instead of StCPGLCP-type · real datasetsset-size std: 0.75 -> 0.78 (CRIME)No clear effect
increase labeled calibration size n from 30 to 500 (ablation)StCP GLCP-type · LogAbs syntheticset-size std gain: 1.12 -> 0.77 at n=30; 0.25 -> 0.18 at n=500Mixed
use StCP on CQR-type at n=30CQR-type · LogAbs syntheticset-size std: 0.98 -> 0.82Improved
use StCP regarding set sizeGLCP-type · CRIME real datamean set size: 3.24 -> 3.11No clear effect
increase lambda (stronger source regularization) (ablation)StCP · LogAbs syntheticset-size stdImproved

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