AI Paper Atlas

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

Unsupervised Few-Shot Image Classification by Learning Features into Clustering Space

Shuo Li, Fang Liu, Zehua Hao, Kaibo Zhao, Licheng Jiao

ECCV 2022 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Embedding adaptation for few-shot classification · 30 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
build tasks from LF2CS clusters instead of clustering-based baselineunsupervised few-shot (ResNet-12) · miniImageNet 5-way 1-shotaccuracy: 48.48 -> 53.14Improved
build tasks from LF2CS clusters instead of CSS-based augmentation viewsunsupervised few-shot (ResNet-12) · miniImageNet 5-way 1-shotaccuracy: 45.42 -> 53.14Improved
build tasks from LF2CS clusters instead of random selectionunsupervised few-shot (ResNet-12) · miniImageNet 5-way 1-shotaccuracy: 29.92 -> 53.14Improved
use LF2CS instead of clustering-based task buildingunsupervised few-shot (Conv-4) · Omniglot 5-way 1-shotaccuracy: 83.14 -> 97.31Improved
use LF2CS instead of clustering-based task buildingunsupervised few-shot (ResNet-12) · tieredImageNet 5-way 1-shotaccuracy: 44.93 -> 53.16Improved
use LF2CS instead of clustering-based task buildingunsupervised few-shot (ResNet-12) · CIFARFS 5-way 1-shotaccuracy: 44.88 -> 55.04Improved
use LF2CS instead of CSS-based task buildingunsupervised few-shot (Conv-4) · CIFARFS 5-way 1-shotaccuracy: 42.59 -> 51.52Improved
use ResNet-12 encoder instead of Conv-4 (ablation)LF2CS · miniImageNet 5-way 1-shotaccuracy: 48.32 -> 53.14Improved
use ResNet-12 encoder instead of Conv-4 (ablation)LF2CS · miniImageNet test KNN top-1accuracy: 49.10 -> 60.25Improved
increase LF2CS head dimension d (ablation)LF2CS · miniImageNet KNN top-1accuracyImproved
evaluate on novel test classes instead of training classes (ablation)LF2CS (Conv-4) · miniImageNet 5-way 1-shotaccuracy: 48.41 -> 48.32No clear effect
evaluate on novel test classes instead of training classes (ablation)LF2CS (ResNet-12) · miniImageNet 5-way 1-shotaccuracy: 53.59 -> 53.14No clear effect

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