AI Paper Atlas

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

Learning From Large-Scale Noisy Web Data With Ubiquitous Reweighting for Image Classification

Jia Li, Yafei Song, Jianfeng Zhu, Lele Cheng, Ying Ming Su, Lin Ye, Pengcheng Yuan, Shumin Han

TPAMI 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 33 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 class reweighting (ablation)baseline ResNeXt-101 · WebVision 2018 valtop-5 accuracy: 71.7% -> 72.8%Improved
add cluster reweighting (ablation)URNet · WebVision 2018 valtop-5 accuracy: 72.8% -> 75.3%Improved
add instance confidence reweighting (ablation)URNet · WebVision 2018 valtop-5 accuracy: 75.3% -> 76%Improved
add bag-specific saliency reweighting (ablation)URNet · WebVision 2018 valtop-5 accuracy: 76% -> 76.3%Improved
add label reweighting (ablation)URNet · WebVision 2018 valtop-5 accuracy: 76.3% -> 76.5%Improved
keep only top 15 Flickr instances per classURNet · WebVision 2018 valtop-5 accuracy: 76.5% -> 77.0%Improved
ensemble 29 modelsURNet · WebVision 2018 valtop-5 accuracy: 77.0% -> 78.3%Improved
add multi-crop testingURNet · WebVision 2018 valtop-5 accuracy: 78.3% -> 79.3%Improved
add multi-scale testingURNet · WebVision 2018 valtop-5 accuracy: 79.3% -> 79.8%Improved

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