AI Paper Atlas

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

Multi-Head Encoding for Extreme Label Classification

Daojun Liang, Haixia Zhang, Dongfeng Yuan, Minggao Zhang

TPAMI 2024 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 2 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 vanilla classifier with MHSResNet-101 classifier · VOC2007 label powersetaccuracy: 79.82 -> 80.34Improved
replace vanilla classifier with MHSResNet-101 classifier · COCO2014 label powersetaccuracy: 46.62 -> 47.25Improved
use MHE instead of AttentionXML HLTXLC classifier · COCO2014 label powersetaccuracy: 46.06 -> 47.21Improved
MHC without label clustering instead of AttentionXMLSingle model · Wiki-500KP@1: 75.01 -> 78.38Improved
longer main head {172;23} instead of {86;46} (ablation)MHC · Eurlex-4KP@1: 85.54 -> 86.00Improved
longer main head {310;43} instead of {155;86} (ablation)MHC · AmazonCat-13KP@1: 95.83 -> 95.94Improved
add MHS pretraining instead of ArcFace-FullResNet-18 · CASIA IJB-CTAR: 94.31 -> 95.06Improved
add MHS pretraining instead of ArcFace-FullResNet-101 · MS1MV2 IJB-CTAR: 96.20 -> 95.85Hurt
add label clustering instead of random decomposition (ablation)small model SimpleConv-4 · generalizationaccuracy: about 4%Improved
add label clustering instead of random decomposition (ablation)overfit ResNet-18 · generalizationaccuracy: the gap between LRD and LC vanishesNo clear effect
use MHP instead of one-hot encoding classifierclassifier · speed at 1 million labelsruntime: 17x faster than OHEImproved
use MHS instead of OHEclassifier · speed at 1 million labelsruntime: 8x faster than OHEImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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