AI Paper Atlas

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

Rethinking Few-shot Image Classification: A Good Embedding is All You Need?

Yonglong Tian, Yue Wang, Dilip Krishnan, Joshua B. Tenenbaum, Phillip Isola

ECCV 2020 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Meta-learning analysis and benchmarks · 1,014 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
use logistic regression instead of nearest neighbour (ablation)ResNet-12 embedding · miniImageNet 5-way 5-shotaccuracy (%): 69.96 -> 78.31Improved
use logistic regression instead of nearest neighbour (ablation)ResNet-12 embedding · miniImageNet 5-way 1-shotaccuracy (%): 56.29 -> 58.74Improved
add L2 feature normalization (ablation)ResNet-12 + LR · miniImageNet 5-way 1-shotaccuracy (%): 58.74 -> 61.56Improved
add 5x support image augmentation (ablation)ResNet-12 + LR + L2 · miniImageNet 5-way 1-shotaccuracy (%): 61.56 -> 62.02No clear effect
add sequential self-distillation of the embedding (ablation)ResNet-12 + LR + L2 + Aug · miniImageNet 5-way 1-shotaccuracy (%): 62.02 -> 64.82Improved
add self-distillation (ablation)Ours · FC100 5-way 5-shotaccuracy (%): 59.13 -> 60.91Improved
train on meta-train + meta-val (trainval) (ablation)ResNet-12 embedding · FC100 5-way 5-shotaccuracy (%): 59.1 -> 66.4Improved
use SEResNet-12 instead of 4-layer ConvNet (ablation)Ours-distill-trainval · miniImageNet 5-way 1-shotaccuracy (%): 56.64 -> 67.73Improved
train embedding with multi-way classification instead of multi-task heads (ablation)ResNet-12 · miniImageNet 5-way 1-shotaccuracy (%): 58.53 -> 62.02Improved
fine-tune on support images at meta-testWRN-28-10 embedding · miniImageNet 5-way 1-shotaccuracy (%): 61.1 -> 57.73Hurt
use self-supervised (MoCo) instead of supervised embeddingResNet50 · miniImageNet 5-way 1-shotaccuracy (%): 57.56 -> 54.19Hurt
add distillation to LR classifier (ablation)ResNet-18 · Meta-Dataset mean accuracyaccuracy (%): 68.02 -> 68.28Improved

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