AI Paper Atlas

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

Rethinking the Dark Knowledge and Kullback-Leibler Divergence Loss in Knowledge Distillation Under Capacity Mismatching

Yingchao Wang, Wenqi Niu, Xingshan Yao, Li You, Weilun Fei

AAAI 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 1 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 cosine similarity instead of MSE on the relative confidence matrix (ablation)RCKD · CIFAR-100 (ResNet-32x4 -> ResNet-8x4)top-1 accuracy: 75.56 -> 76.89Improved
use cosine similarity instead of MSE on the relative confidence matrix (ablation)RCKD · CIFAR-100 (WRN-40-2 -> ShuffleNet-V1)top-1 accuracy: 76.13 -> 76.97Improved
use cosine similarity instead of MSE on the relative confidence matrix (ablation)RCKD · CIFAR-100 (VGG-13 -> MobileNet-V2)top-1 accuracy: 68.33 -> 69.77Improved
replace KL divergence distillation with RCKDKD · CIFAR-100 (ResNet-32x4 -> ResNet-8x4)top-1 accuracy: 73.33 -> 76.89Improved
replace KL divergence distillation with RCKDKD · ImageNet (ResNet-34 -> ResNet-18)top-1 accuracy: 71.03 -> 72.02Improved
replace KL divergence distillation with RCKDKD · CIFAR-100 (VGG-13 -> MobileNet-V2)top-1 accuracy: 67.37 -> 69.77Improved
use RCKD instead of the best logit-based baselinelogit-based KD · CIFAR-100 (ResNet-50 -> MobileNet-V2)top-1 accuracy: 70.35 -> 69.67Hurt
use RCKD instead of ReviewKD feature-based distillationfeature-based KD · ImageNet (ResNet-50 -> MobileNet-V1)top-1 accuracy: 72.56 -> 72.41Hurt
vary the RCKD weight beta (too small or too large) (ablation)RCKD · CIFAR-100top-1 accuracyMixed

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