AI Paper Atlas

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

LLM-assisted Entropy-based Adaptive Distillation for Unsupervised Fine-grained Visual Representation Learning

Jianfeng Dong, Danfeng Luo, Daizong Liu, Jie Sun, Xiaoye Qu, Xun Yang, Dongsheng Liu, Xun Wang

ICCV 2025 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 2 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 CLIP knowledge distillation to MoCo (ablation)MoCo · Cub classificationTop-1: 62.91 -> 75.46Improved
add contrastive learning to distillation (ablation)LEAD · Cub classificationTop-1: 75.46 -> 76.03Improved
add entropy-based adaptive weighting (ablation)LEAD (CL+KD) · Cub classificationTop-1: 76.03 -> 78.44Improved
add entropy-based adaptive weighting (ablation)LEAD (CL+KD) · Cub retrievalmAP: 38.87 -> 44.43Improved
use LLM descriptions instead of fixed template (ablation)LEAD · Cub classificationTop-1: 76.63 -> 78.44Improved
use LLM descriptions instead of fixed template (ablation)LEAD · Car classificationTop-1: 85.94 -> 88.14Improved
use LLM descriptions instead of fixed template (ablation)LEAD · Aircraft classificationTop-1: 64.81 -> 67.45Improved
use LLM-assisted distillation instead of feature-based distillation (ablation)LEAD · Cub retrievalmAP: 28.64 -> 44.43Improved
use LLM-assisted distillation instead of relation-based distillation (ablation)LEAD · Car classificationTop-1: 75.68 -> 88.14Improved
use information entropy instead of maximum probability for confidence (ablation)LEAD · Aircraft classificationTop-1: 63.67 -> 67.45Improved
use information entropy instead of maximum probability for confidence (ablation)LEAD · Cub classificationTop-1: 78.07 -> 78.44Improved
replace dataset categories with LLM-generated categories (ablation)LEAD · Cub classificationTop-1: 78.44 -> 78.20No 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