AI Paper Atlas

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

QuEST: Quantized Embedding Space for Transferring Knowledge

Himalaya Jain, Spyros Gidaris, Nikos Komodakis, Patrick Pérez, Matthieu Cord

ECCV 2020 · Label-efficient and robust learning › Knowledge distillation · approach: Few-shot and metric distillation · 12 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
distill on last feature level (Block L) instead of previous stage (Block L-1) (ablation)QuEST · CIFAR-100 WRN-40-2 -> WRN-16-2top-1 accuracy: 74.16 -> 76.10Improved
distill on last feature level (Block L) instead of previous stage (Block L-1) (ablation)QuEST · CIFAR-10 WRN-40-1 -> WRN-16-1top-1 accuracy: 91.64 -> 91.98Improved
predict visual word assignments instead of regressing raw features (ablation)QuEST vs feature regression · CIFAR-100 ResNet50 -> VGG8top-1 accuracy: 71.9 -> 75.17Improved
predict visual word assignments instead of regressing raw features (ablation)QuEST vs feature regression · CIFAR-100 WRN-40-2 -> WRN-16-2top-1 accuracy: 75.53 -> 76.10Improved
predict visual word assignments instead of regressing raw features, same-architecture teacher (ablation)QuEST vs feature regression · CIFAR-100 WRN40-2 -> WRN40-2top-1 accuracy: 78.26 -> 77.96Hurt
predict visual word assignments instead of regressing raw features, same-architecture teacher (ablation)QuEST vs feature regression · CIFAR-100 ResNet32x4 -> ResNet32x4top-1 accuracy: 80.57 -> 80.53No clear effect
use peaky soft assignment instead of hard assignment (tau=0) (ablation)QuEST · CIFAR-100top-1 accuracyImproved
use small temperature tau (peaky soft assignment) rather than large tau (ablation)QuEST · CIFAR-100top-1 accuracyImproved
increase vocabulary size K to at least 2048 (ablation)QuEST · CIFAR-100top-1 accuracyNo clear effect
add the KD loss on top of QuESTQuEST · CIFAR-100 / ImageNettop-1 accuracyNo clear effect
QuEST instead of CRD+KD, ResNet18 studentQuEST · ImageNet val ResNet34 -> ResNet18top-1 error: 28.62 -> 28.33Improved
QuEST instead of OFD, MobileNet studentQuEST · ImageNet val ResNet50 -> MobileNettop-1 error: 28.75 -> 27.46Improved

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