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

Prototype-driven contrastive learning enables transferable seismic representation

delete2026-06-10
delete0
PRE
AI
F
Fanchun Meng
T
Tao Ren
H
Hengwei Zhang
X
Xinyue Wang *
DOI：10.1016/j.engappai.2026.115317delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
• ConProSeis learns transferable representations from unlabeled seismic waveforms. • Physics-guided augmentations preserve waveform semantics under perturbations. • Prototype-level contrast reduces semantic shifts from strong perturbations. • An EMA dual-encoder stabilizes prototype evolution against strong perturbations. • Learned representations transfer well to classification, regression, and detection.

Journal

Engineering Applications of Artificial Intelligence cover
Engineering Applications of Artificial Intelligence
IF:
8
Papers:
5.3K
Citations:
3.5W

Organization

No organization information available