AI Paper Atlas

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

Maximum-Margin Hamming Hashing

Rong Kang, Yue Cao, Mingsheng Long, Jianmin Wang, Philip S. Yu

ICCV 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 45 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 quantization loss (ablation)MMHH · NUS-WIDE, MS-COCO, CIFAR-10MAP@H<=2: 4.0%, 3.1% and 4.2%Improved
use cosine distance instead of Euclidean distance as Hamming proxy (ablation)MMHH · NUS-WIDE, MS-COCO, CIFAR-10MAP@H<=2: 16.2%, 12.2% and 22.8%Improved
add semi-batch optimization (ablation)MMHH, DCH, HashNet · ImageNet subsets (100-1000 classes)MAP@H<=2Improved
more classes increase the gain from semi-batch optimizationMMHH · ImageNet-1000 vs ImageNet-100MAP@H<=2Improved
replace Cauchy loss (DCH) with max-margin t-distribution lossDCH · NUS-WIDE 50% label noiseMAP drop: 0.06 -> 0.02Improved
replace sigmoid loss (HashNet) with max-margin t-distribution lossHashNet · NUS-WIDE 50% label noiseMAP drop: 0.17 -> 0.02Improved
use max-margin t-distribution loss instead of HashNet sigmoid lossHashNet · MS-COCO 48 bitsqueries returning no images: 44% -> 13%Improved
use max-margin loss instead of Cauchy loss to reduce loss on in-ball similar pairsDCH · MS-COCO 48 bitsshare of total loss from retrievable similar pairs: ~40%Improved
hash lookup instead of linear scanMMHH · MS-COCOsearch time per query: 10.2 ms/query -> 0.9 ms/queryImproved
use deep features with shallow SDH vs end-to-end MMHHMMHH vs SDH · NUS-WIDE, MS-COCO, CIFAR-10average MAP@H<=2: 17.5%, 19.0% and 23.5%Improved
evaluate on unseen classes instead of seen classesMMHH · MS-COCO unseen-class protocolMAPHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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