AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Rebalancing Using Estimated Class Distribution for Imbalanced Semi-Supervised Learning under Class Distribution Mismatch

Taemin Park, Hyuck Lee, Heeyoung Kim

ECCV 2024 · Label-efficient and robust learning › Semi-supervised learning · approach: Confidence-based pseudo-label selection · 3 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
replace ABC's labeled-distribution-based mask with RECD's estimated unlabeled class distributionReMixMatch+ABC · CIFAR-10-LT (gamma_l = 100, gamma_u = 1)bACC: 76.4 -> 90.3Improved
replace ABC's mask with RECD estimated class distribution (FixMatch)FixMatch+ABC · CIFAR-10-LT (gamma_l = 100, gamma_u = 1)bACC: 82.7 -> 90.2Improved
add RECD under matched class distributionsFixMatch+ABC · CIFAR-10-LT (gamma = 100)bACC: 81.1 -> 84.0Improved
add RECD under matched class distributions with ReMixMatch backboneReMixMatch+ABC · CIFAR-10-LT (gamma = 100)bACC: 84.5 -> 85.4No clear effect
add logit adjustment (LA) to SAW under severe class distribution mismatchReMixMatch*+SAW · CIFAR-10-LT (gamma_l = 100, gamma_u = 1)bACC: 87.0 -> 74.2Hurt
add adaptively adjusted feature multiplier (AAFM) to RECD (ablation)FixMatch+RECD · CIFAR-10-LT (gamma_l = 100, gamma_u = 1)minority-class test lossImproved
estimate unlabeled class distribution with EMA of predictions in RECDRECD · CIFAR-10-LT (gamma_l = 100, gamma_u = 1)estimated versus true class distributionImproved

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