AI Paper Atlas

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

Active Label Correction Using Robust Parameter Update and Entropy Propagation

Kwang In Kim

ECCV 2022 · Label-efficient and robust learning › Noisy labels · approach: Noise-aware loss and pairwise methods

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 robust contribution-weighted parameter update instead of uniform gradient combination (ablation)active label correction · CIFAR-100accuracyImproved
use soft robust update instead of hard noise detection [14,43] (ablation)active label correction · CIFAR-100accuracyImproved
add entropy diffusion to batch selection (ablation)active label correction · Caltech-256accuracyImproved
add entropy diffusion versus conventional batch entropy selection (pure label acquisition) (ablation)entropy-based selection · CIFAR-100accuracyImproved
use the proposed method instead of DALC and CorrActiveactive label correction · CIFAR-10 (uniform noise)labels to reach 90% accuracy: 9,000 / 11,000 -> 3,000Improved
use the proposed method instead of loss-based active cleaning [2] for noisy sample selectionactive label correction · CIFAR-100noisy example selection accuracy: 91.1 -> 97.9Improved
use LossEnt selection instead of RandomLossEnt · CIFAR-10 and F-MNISTaccuracyHurt
use Entropy or CorrActive instead of Randomactive label correction · benchmarksaccuracyImproved
add class-transition noise model learning (DALC) to entropy selectionactive label correction · benchmarksaccuracyImproved

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