logo
CanyamAI summaries for academic research
language
MobileMobile
  • Phone FeatureWeChat Mini Program
EnglishEnglish
noticeNotice
BlogBlog
delete
arrow
Return

Gradient amplification for gradient matching based dataset distillation

delete2025-07-05
delete0
PRE
AI
J
Jingxuan Zhang
Z
Zhihua Chen
L
Lei Dai
李平 cover
李平 (Ping Li)
盛
盛斌 (Bin Sheng)
DOI：10.1016/j.neunet.2025.107819delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
• Label cycle shifting strategy produces informative gradient information. • The early exit mechanism alleviates matching the useless gradients. • Ensembling distilled datasets makes the training process more stable. • Gradient matching and distribution matching mutually enhance each other.

Journal

Neural Networks cover
Neural Networks
IF:
6.3
Papers:
7.8K
Citations:
3.0W

Organization

No organization information available