AI Paper Atlas

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

Distilling Virtual Examples for Long-Tailed Recognition

Yin-Yin He, Jianxin Wu, Xiu-Shen Wei

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Decoupled and calibrated long-tail training · 117 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 BSCE loss term to the distillation loss (ablation)DiVE · CIFAR-100-LT (imbalance 100)top-1 accuracy: 44.50 -> 45.35Improved
add power normalization to virtual examples (tau=1, label-mixed teacher) (ablation)DiVE variant · CIFAR-100-LT (imbalance 100)top-1 accuracy: 41.59 -> 44.55Improved
raise temperature from 1 to 3 (on power-normalized label-mixed teacher) (ablation)DiVE variant · CIFAR-100-LT (imbalance 100)top-1 accuracy: 44.55 -> 43.22Hurt
raise temperature to 3 and use power normalization on tau-softened teacher without ground truth (ablation)DiVE without BSCE · CIFAR-100-LT (imbalance 100)top-1 accuracy: 43.25 -> 44.50Improved
replace label-mixed teacher with plain teacher logits (tau=3, power) (ablation)DiVE without BSCE · CIFAR-100-LT (imbalance 100)top-1 accuracy: 43.22 -> 44.50Mixed
distill virtual examples from a model trained on full CIFAR-100 (tau=3) instead of CE teacherstudent · CIFAR-100-LT (imbalance 100)top-1 accuracy: 39.20 -> 61.58Improved
use a more balanced teacher (BSCE instead of CE) for distillationstudent · CIFAR-100-LT (imbalance 100)top-1 accuracy: 39.20 -> 43.25Improved
apply DiVE on top of BSCEBSCE · ImageNet-LTtop-1 accuracy: 50.48 -> 53.10Improved
apply DiVE on top of BSCEBSCE · iNaturalist2018 (90 epochs)top-1 accuracy: 65.35 -> 69.13Improved
train DiVE for 200 instead of 90 epochsDiVE · iNaturalist2018top-1 accuracy: 69.13 -> 71.71Improved
distill RIDE with DiVE (RIDE-DiVE)RIDE · iNaturalist2018top-1 accuracy: 72.60 -> 73.44Improved
distill RIDE with DiVE (RIDE-DiVE)RIDE · CIFAR-100-LT (imbalance 100)top-1 accuracy: 49.10 -> 51.66Improved

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