AI Paper Atlas

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

VRM: Knowledge Distillation via Virtual Relation Matching

Weijia Zhang, Fei Xie, Weidong Cai, Chao Ma

ICCV 2025 · Label-efficient and robust learning › Knowledge distillation · approach: Distillation with causal and adaptive guidance · 9 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 vanilla KD loss with VRM virtual relation matchingDeiT-T student with ResNet56 teacher · CIFAR-100 cross-architecturetop-1 accuracy: 65.08 -> 79.52Improved
use VRM relation matching instead of DIST on same-model KDResNet32x4 -> ResNet8x4 · CIFAR-100top-1 accuracy: 76.31 -> 78.76Improved
supervise student virtual-view predictions with ground-truth CE (ablation)VRM · CIFAR-100 ResNet32x4 -> ResNet8x4top-1 accuracy: 78.47 -> 78.76Improved
supervise student virtual-view predictions with ground-truth CE (ablation)VRM · CIFAR-100 VGG13 -> VGG8top-1 accuracy: 75.66 -> 76.19Improved
add vertex matching to edge matching (ablation)VRM · CIFAR-100top-1 accuracyHurt
train for 360 epochs instead of 240 (ablation)VRM · CIFAR-100top-1 accuracy: +0.21%Improved
use class-wise pairwise-difference relations instead of Gram matrices or angle-wise relations (ablation)VRM · CIFAR-100top-1 accuracyImproved
prune redundant edges (REP) from the augmented graph (ablation)VRM · CIFAR-100top-1 accuracy and efficiencyImproved
prune unreliable edges (UEP) by joint entropy (ablation)VRM · CIFAR-100top-1 accuracyImproved
use weak-only transformations for both real and virtual views (ablation)VRM · CIFAR-100 / ImageNettop-1 accuracyNo clear effect
increase batch size (ablation)DIST relation matching versus VRM · CIFAR-100top-1 accuracyMixed
apply VRM to object detection versus feature-based FCFDFaster-RCNN ResNet18 student · MS-COCOAP: 37.37 -> 35.46 (FCFD vs VRM)Hurt

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