AI Paper Atlas

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

Diversity-Enhanced Distribution Alignment for Dataset Distillation

Hongcheng Li, Yucan Zhou, Xiaoyan Gu, Bo Li, Weiping Wang

ICCV 2025 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias · 4 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 class-specific mean alignment from stored Gaussian statistics in place of BN alignment (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracy: 47.5 -> 63.1Improved
add covariance alignment to mean alignment (ablation)DEDA · CIFAR-100 IPC=10top-1 accuracy: 39.8 -> 41.6Improved
add covariance alignment to mean alignment (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracy: 63.1 -> 63.8Improved
add last-layer variance maximization regularizer LVar (ablation)DEDA · CIFAR-100top-1 accuracy: +0.6% and 0.9%Improved
add off-diagonal covariance minimization LCov (ablation)DEDA · CIFAR-100top-1 accuracy: +1.4% and 0.4%Improved
add LCov without LVar (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracy: 63.8 -> 64.2Improved
increase covariance weight gamma from 0 to 50 (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracy: 0 -> 50Improved
increase gamma beyond 50 to 200 (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracy: 50 -> 200Hurt
use large lambda1 for variance regularization (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracyHurt
vary lambda2 covariance regularization weight (ablation)DEDA · CIFAR-100 IPC=50top-1 accuracyNo clear effect
replace SRe2L BN alignment with DEDA, ResNet-18 distillationDEDA vs SRe2L · CIFAR-10 IPC=50top-1 accuracy: 47.5 -> 60.2Improved
replace SRe2L with DEDA on ResNet-50 cross-architectureDEDA vs SRe2L · CIFAR-100 IPC=50 cross-archtop-1 accuracy: 52.8 -> 63.7Improved

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