AI Paper Atlas

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

Local Dense Logit Relations for Enhanced Knowledge Distillation

Liuchi Xu, Kang Liu, Jinshuai Liu, Lu Wang, Lisheng Xu, Jun Cheng

ICCV 2025 · Label-efficient and robust learning › Knowledge distillation · approach: Distillation with causal and adaptive guidance · 15 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 LDRLD instead of vanilla KDResNet32x4->ResNet8x4 · CIFAR-100 valtop-1 accuracy: 73.33 -> 77.20Improved
use LDRLD instead of vanilla KDResNet34->ResNet18 · ImageNet-1K valtop-1 accuracy: 70.66 -> 71.88Improved
use LDRLD instead of WTTMResNet34->ResNet18 · ImageNet-1K valtop-1 accuracy: 72.19 -> 71.88Hurt
add Adaptive Decay Weight (IRW and ERD) to LDRLD (ablation)VGG13->VGG8 · CIFAR-100top-1 accuracy: 74.25 -> 74.46Improved
add Adaptive Decay Weight (IRW and ERD) to LDRLD (ablation)ResNet50->MobileNetV2 · CIFAR-100top-1 accuracy: 69.23 -> 69.52Improved
add remaining non-target knowledge loss L_RNTK to L_Local (ablation)ResNet32x4->ResNet8x4 · CIFAR-100top-1 accuracy: 76.55 -> 77.20Improved
add local relational loss L_Local to L_RNTK (ablation)ResNet32x4->ResNet8x4 · CIFAR-100top-1 accuracy: 75.78 -> 77.20Improved
use recursion depth d=7 instead of smaller or larger depths (ablation)LDRLD · CIFAR-100top-1 accuracyImproved
use RNTK (d=7) instead of NCKD for non-target knowledge (ablation)VGG13->VGG8 · CIFAR-100top-1 accuracy: 74.15 -> 74.28Improved
train with 0.1 noisy labels (ablation)LDRLD · CIFAR-100top-1 accuracy: 77.20 -> 77.32No clear effect
combine LDRLD with DKDResNet32x4->ResNet8x4 · CIFAR-100top-1 accuracy: 77.20 -> 77.42Improved
combine LDRLD with IKDResNet32x4->ShuffleNetV2 · CIFAR-100top-1 accuracy: 77.33 -> 77.62Improved

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