AI Paper Atlas

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

Probabilistic Knowledge Distillation of Face Ensembles

Jianqing Xu, Shen Li, Ailin Deng, Miao Xiong, Jiaying Wu, Jiaxiang Wu, Shouhong Ding, Bryan Hooi

CVPR 2023 · Label-efficient and robust learning › Knowledge distillation · approach: Patient consistent distillation · 8 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 mean averaging with Bayesian, confidence-weighted ensemble averagingResNet34 face ensemble · IJB-BTAR at FAR 1e-5: 86.0 -> 89.7Improved
distill ensemble feature embeddings and BEA confidence statisticResNet12 student · IJB-BTAR at FAR 1e-5: 81.64 vs 77.96Improved
distill ensemble feature embeddings and BEA confidence statisticMobileFaceNet student · IJB-BTAR at FAR 1e-5: 83.45 vs 81.85Improved
add Bayesian ensemble averaging rather than mean ensembleResNet34 face ensemble · CFP-FPaccuracy: 98.8 -> 99.0Improved
add Bayesian ensemble averaging rather than mean ensembleResNet34 face ensemble · CPLFWaccuracy: 93.7 -> 94.3Improved
distill from one deterministic teacher rather than BEA (ablation)ResNet12 student · IJB-BTAR at FAR 1e-5: 81.64 -> 77.91Hurt
distill from one SCF-trained teacher rather than BEA (ablation)ResNet12 student · IJB-BTAR at FAR 1e-5: 81.64 -> 80.55Hurt
distill from a deterministic teacher ensemble rather than BEA (ablation)ResNet12 student · IJB-BTAR at FAR 1e-5: 81.64 -> 78.92Hurt
distill from SCF-trained ensemble teachers rather than BEA (ablation)ResNet12 student · IJB-BTAR at FAR 1e-5: 81.64 -> 81.01Hurt
distill from one SCF teacher rather than BEA (ablation)ResNet12 student · IJB-BTAR at FAR 1e-5: 81.64 -> 80.93Hurt
add the BEA confidence statistic to weighted probabilistic ensemble distillationface-recognition student models · IJB-BTAR at FAR 1e-5Improved

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