AI Paper Atlas

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

Learning From Noisy Web Data With Category-Level Supervision

Li Niu, Qingtao Tang, Ashok Veeraraghavan, Ashutosh Sabharwal

CVPR 2018 · Label-efficient and robust learning › Noisy labels · approach: Instance-dependent noise modeling · 26 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 category-level hybrid semantic supervision and jointly train the classifier with semantic VAEWSCI · AwA2, CUB, and SUN Attributeaccuracy (%): 86.41 -> 91.14; 75.02 -> 77.34; 40.56 -> 42.26Improved
Add category-level information using a semantic VAE, rather than a detached plain VAE without category-level information (ablation)WSCI · AwA2, CUB, and SUN Attributeaccuracy (%): 86.15 -> 88.88; 74.17 -> 76.28; 37.91 -> 41.23Improved
Jointly leverage category-level supervision in the classifier and semantic VAE, rather than use separate classification flows (ablation)WSCI · AwA2, CUB, and SUN Attributeaccuracy (%): 88.88 -> 91.14; 76.28 -> 77.34; 41.23 -> 42.26Improved
Add visual encoding to the hybrid semantic representation (ablation)WSCI · AwA2, CUB, and SUN Attributeaccuracy (%): 90.52 -> 91.14; 76.86 -> 77.34; 41.97 -> 42.26Improved
Add attribute vectors to the hybrid semantic representation (ablation)WSCI · AwA2, CUB, and SUN Attributeaccuracy (%): 89.56 -> 91.14; 76.61 -> 77.34; 41.03 -> 42.26Improved
Use WSCI rather than Xiao et al.SUN Attribute · s(201) and s(401)accuracy (%): 39.85 -> 41.79; 39.00 -> 41.47Improved

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