arrow
返回

Learning Dynamic Batch-Graph Representation for Deep Representation Learning

delete2024-07-19
delete0
PRE
AI
X
Xixi Wang
B
Bo Jiang *
X
Xiao Wang
B
Bin Luo
DOI:10.1007/s11263-024-02175-8delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
Recently, batch-based image data representation has been demonstrated to be effective for context-enhanced image representation. The core issue for this task is capturing the dependences of image samples within each mini-batch and conducting message communication among different samples. Existing approaches mainly adopt self-attention or local self-attention models (on patch dimension) for this task which fail to fully exploit the intrinsic relationships of samples within mini-batch and also be sensitive to noises and outliers. To address this issue, in this paper, we propose a flexible Dynamic Batch-Graph Representation (DyBGR) model, to automatically explore the intrinsic relationship of samples for contextual sample representation. Specifically, DyBGR first represents the mini-batch with a graph (termed batch-graph) in which nodes represent image samples and edges encode the dependences of images. This graph is dynamically learned with the constraint of similarity, sparseness and semantic correlation. Upon this, DyBGR exchanges the sample (node) information on the batch-graph to update each node representation. Note that, both batch-graph learning and information propagation are jointly optimized to boost their respective performance. Furthermore, in practical, DyBGR model can be implemented via a simple plug-and-play block (named DyBGR block) which thus can be potentially integrated into any mini-batch based deep representation learning schemes. Extensive experiments on deep metric learning tasks demonstrate the effectiveness of DyBGR. We will release the code at https://github.com/SissiW/DyBGR.
Keyword:
Image representation learning
Transformer
Graph learning
Metric learning
Mini-batch

期刊

International Journal of Computer Vision 封面图
International Journal of Computer Vision
IF:
9.3
论文数:
3.9K
被引数:
2.8W

机构

A
anhui university
学者数:
1.9W
论文数: 1.2W
被引数: 24
引用论文

引用论文

Profiles of Gene Expression in Human Autoimmune Disease
err2004-01-01
err0
PREAI
errThomas M. Aune; Kevin Maas; Joel Parker; Jason H. Moore; Nancy J. Olsen
err分享
err收藏
A case of Shwachman‐Diamond syndrome in a male neonate
err2007-01-02
err0
PREAI
errMirjana Todorovič‐Guid; Olga Krajnc; Natasa Marcun Varda; Boris Zagradišnik; Nadja Kokalj Vokac; Alojz Gregoric
err分享
err收藏
Radix Scrophulariae extracts (harpagoside) suppresses hypoxia-induced microglial activation and neurotoxicity
err2015-09-14
err0
errOAAI
errShiow-Yunn Sheu; Yi-Wen Hong; Jui-Sheng Sun; Man-Hai Liu; Ching-Yun Chen; Cherng-Jyh Ke
err分享
err收藏
err分享
err收藏
Evolution of structural and optical properties of photocatalytic Fe doped TiO2 thin films prepared by RF magnetron sputtering
err2014-01-01
err0
PREAI
errPrabitha B. Nair; L. V. Maneeshya; V. B. Justinvictor; Georgi P. Daniel; K. Joy; P. V. Thomas
err分享
err收藏
学者 查看更多内容