AI Paper Atlas

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

Multi-label Learning with Missing Labels Using Mixed Dependency Graphs

Baoyuan Wu, Fan Jia, Wei Liu, Bernard Ghanem, Siwei Lyu

IJCV 2018 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 97 citations (OpenAlex)

Read the paper

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 class co-occurrence graph with sparse and low rank decomposition (95% missing, original init)MLMG · MediaMill, ESP Game, IAPRTC-12AP relative improvement: 2.43%, 6.29%, 3.94%, 2.78%, 4.7%Improved
replace class co-occurrence graph with sparse and low rank decomposition (95% missing, filled init)MLMG · MediaMill, ESP Game, IAPRTC-12AP relative improvement: 6.38%, 0.23%, -0.87%, 2.76%, 1.43%Mixed
use MLMG-CO instead of MLMG-SL when missing label proportion is smallMLMG · MediaMill, ESP Game, IAPRTC-12AP and mAPMixed
fill initial label matrix with ancestor labels from the semantic hierarchy (ablation)MLMG · four multi-label datasetsAP and mAPImproved
add semantic hierarchical constraint to MLMG-CO (ablation)MLMG-CO · four multi-label datasetsAHL: positive (filling only) -> 0Improved
add both filling and constraint to MLMG-CO (ablation)MLMG-CO · four multi-label datasetsAP and mAPImproved
use semantic hierarchy as constraint during optimization instead of CSSAG post-processingMLMG-CO · four multi-label datasetsmAPImproved
use MLMG-CO/SL without hierarchy information compared to best competitorMLMG · six datasetsAP and mAP margin: at least 5% (AP) or 3% (mAP)Improved
compute PGD step size once every 5 iterations with damping 0.9 (ablation)MLMG-CO · runtimeruntime: 50% -> about 40%Improved
initialize missing labels randomly instead of zero (ablation)MLMG-CO and MLMG-SL · 50% missing labelsAPNo clear effect
generate missing labels on all classes instead of only leaf and singleton classes (ablation)MLMG-CO and MLMG-SL · training imputationAP and mAPHurt
switch from traditional to CNN features (ESP Game testing AP, MLMG-CO)MLMG-CO · ESP Game semi-supervisedAP: 0.4422 -> 0.5887Improved

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