[PDF][PDF] ISTANBUL TECHNICAL UNIVERSITY 击GRADUATE SCHOOL

HDB SELF-ATTENTION - 2023 - polen.itu.edu.tr
… We paid attention to an approach that considers the high number of users and items of
real-world e-commerce platforms and can work reasonably well in this case. For dealing with the …

[HTML][HTML] 基于深度学习的跨模态检索综述

尹奇跃, 黄岩, 张俊格, 吴书, 王亮 - 2021 - cjig.cn
… Accordingly, cross-modal retrieval has attracted considerable attention. One type of data is
… is formed as regularizers or loss functions to enforce a pairwise constraint for learning multi-…

跨模态数据实体分辨研究综述

曹建军, 聂子博, 郑奇斌, 吕国俊, 曾志贤 - 软件学报, 2023 - jos.org.cn
… 基于图正则化的方法 图正则化(graph regularization), 通过探索图中顶点和边丰富的关系,
能够… block attention module, CBAM), 能够在利用CNN 方法提取图像特征时, 通过计算通道(channel) …

跨媒体智能关联分析与语义理解理论与技术研究进展

于俊清, 王鑫, 况琨, 刘偲, 张新峰, 宋子恺 - 计算机辅助设计与图形学学报, 2023 - jcad.cn
… Yang 等[128]提出了一种堆叠式注意网络 (stacked attention networks, SAN), 利用文本问题 的
语义… Semi-supervised cross-media feature learning with unified patch graph regularization[J]. …

基于深度学习的MRI 重建方法综述.

邓戈文, 魏国辉, 马志庆 - Journal of Computer Engineering …, 2023 - search.ebscohost.com
… 通过设计可变形注意力(deformable attention)来 揭示模型的可解释… 比如MoDL-SToRM(MoDL
with smoothness regularization on … auto-encoders[J].Medical Physics, 2018,45(5):2023-2032. …