AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Impact of base dataset design on few-shot image classification

Othman Sbai, Camille Couprie, Mathieu Aubry

ECCV 2020 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Embedding adaptation for few-shot classification · 27 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
train on full MiniIN6K (6000 classes) instead of the 64 MiniIN base classesCosine Classifier, WRN28-10 · MiniIN test 5-shotaccuracy: 78.95 -> 96.91Improved
train on full MiniIN6K instead of MiniIN base classesCosine Classifier, WRN28-10 · MiniIN test 1-shotaccuracy: 61.62 -> 85.40Improved
use random 64 classes from MiniIN6K instead of the MiniIN benchmark base classesCosine Classifier, WRN28-10 · MiniIN test 5-shotaccuracy: 78.95 -> 75.48Hurt
train on full MiniIN6K* instead of CUB base classesCosine Classifier, WRN28-10 · CUB test 5-shotaccuracy: 90.32 -> 90.89No clear effect
train ProtoNet on full MiniIN6K instead of MiniIN baseProtoNet, WRN28-10 · MiniIN test 5-shotaccuracy: 73.64 -> 85.14Improved
train on IN6k instead of the 389 ImageNet base classesCosine Classifier, ResNet-34 · ImageNet benchmark 5-shottop-5 accuracy: 78.15 -> 88.50Improved
use about 384 classes with 100 images per class instead of 64 classes with 600 images for a fixed budget (ablation)Cosine Classifier · MiniIN test 5-shotaccuracy: around 5%Improved
split base classes into 8 sub-classes using oracle features (ablation)Cosine Classifier, ResNet-34 · ImageNet benchmark 5-shottop-5 accuracy: 78.15 -> 80.30Improved
split base classes into 4 sub-classes using oracle features (ablation)Cosine Classifier, ResNet-34 · ImageNet benchmark 1-shottop-5 accuracy: 57.56 -> 58.49Improved
select base classes closest rather than farthest to the test classes (ablation)Cosine Classifier, WRN28-10 · MiniIN test 5-shotaccuracyImproved
regroup or split classes toward the optimal class count (grouping if too many classes, splitting if too few) (ablation)Cosine Classifier · MiniIN and CUB testaccuracyMixed
use balanced bisection/pairing relabeling instead of K-means (ablation)Cosine Classifier · MiniIN testaccuracyImproved

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