AI Paper Atlas

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

Distill on a Diet: Efficient Knowledge Distillation via Learnable Data Pruning

Yifan Wu, Yiqi Wang, Xichen Ye, Wenjing Yan, Xiaoqiang Li, Cheng Jin, WEIZHONG ZHANG, Xiangyu Yue

ECCV 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Distillation with causal and adaptive guidance · 0 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
replace fixed-threshold CCS selection with learnable Beta policy (ablation)IF-CCS · CIFAR-100 KD (prune 50%)accuracy: 77.64 -> 79.10Improved
replace fixed-window BWS selection with learnable Beta policy (ablation)IF-BWS · CIFAR-10 KD (prune 90%)accuracy: 86.20 -> 88.51Improved
replace fixed-window BWS selection with learnable Beta policy (ablation)IF-BWS · CIFAR-100 KD (prune 90%)accuracy: 58.71 -> 64.11Improved
replace fixed-window BWS selection with learnable Beta policy (ablation)IF-BWS · ImageNet KD (prune 50%)accuracy: 73.63 -> 74.02Improved
prune 30% of CIFAR-10 with IF-Beta instead of full dataIF-Beta · CIFAR-10 KDaccuracy: 95.50 -> 95.69Improved
prune data with IF-Beta instead of full dataIF-Beta · CIFAR-100 KDaccuracy: 79.38 -> 79.53Improved
prune 50% of ImageNet with IF-Beta instead of full dataIF-Beta · ImageNet KDaccuracy: 73.54 -> 74.02Improved
select data with IF-Beta instead of random pruning under limited budgetIF-Beta · CIFAR-100 KD limited budget (prune 50%)accuracy: 63.39 -> 67.31Improved
use generic ImageNet-pretrained model instead of in-domain model for influence scores (no KD) (ablation)IF-Beta · CIFAR-100 standard pruning (prune 90%)accuracy: 55.78 -> 53.31Hurt
use IF-Beta instead of Medium-Difficulty* for efficient KDIF-Beta · ImageNet KD ResNet-34 to MobileNetaccuracy: 70.92 -> 71.20Improved
use IF-Beta instead of full-data baseline for training timeIF-Beta · ImageNet KD ResNet-34 to MobileNettraining time (h): 39.86 -> 19.24Improved
use IF-Beta instead of random pruning with FitNets lossIF-Beta · CIFAR-100 FitNets KD (prune 90%)accuracy: 64.99 -> 71.52Improved

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