AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Asymmetric Metric Learning for Knowledge Transfer

Mateusz Budnik, Yannis Avrithis

CVPR 2021 · Label-efficient and robust learning › Knowledge distillation · approach: Few-shot and metric distillation · 30 citations (OpenAlex)

Check 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
Use asymmetric metric learning with contrastive loss rather than symmetric student-only contrastive trainingMobileNetV2 · R-Oxford5k medium, symmetric; VGG16 teachermAP: 53.6 -> 57.3Improved
Add the teacher's anchor representation as a positive for the student anchor (ablation)Contrastive asymmetric training · R-Oxford5k medium, symmetric; MobileNetV2, VGG16 teachermAP: 57.3 -> 57.3Improved
Add negatives to contrastive asymmetric training that uses only positive pairs (ablation)Contrastive asymmetric training · R-Oxford5k medium, symmetric; MobileNetV2, VGG16 teachermAP: 55.9 -> 57.3Improved
Combine self-positive regression with labeled contrastive metric learning (ablation)MobileNetV2 · R-Oxford5k medium, asymmetric; VGG16 teachermAP: 48.0 -> 42.9Hurt
Use plain feature regression instead of relational distillationMobileNetV2, VGG16 teacher · R-Oxford5k medium, asymmetricmAP: 2.0 -> 48.0Improved
Use plain feature regression instead of DarkRankMobileNetV2, VGG16 teacher · R-Oxford5k medium, asymmetricmAP: 1.7 -> 48.0Improved
Use Contr+ rather than plain contrastive training (ablation)MobileNetV2, ResNet101 teacher · R-Oxford5k medium, asymmetricmAP: 32.3 -> 47.1Improved
Use asymmetric metric learning rather than student-only trainingEfficientNet-B3, ResNet101 teacher · R-Paris6k medium, symmetricmAP: 75.1 -> 77.1Mixed

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