典型文献
Multi-attention fusion and weighted class representation for few-shot classification
文献摘要：
The existing few-shot learning ( FSL) approaches based on metric-learning usually lack atten- tion to the distinction of feature contributions, and the importance of each sample is often ignored when obtaining the class representation, where the performance of the model is limited. Additional- ly, similarity metric method is also worthy of attention. Therefore, a few-shot learning approach called MWNet based on multi-attention fusion and weighted class representation ( WCR) is proposed in this paper. Firstly, a multi-attention fusion module is introduced into the model to highlight the valuable part of the feature and reduce the interference of irrelevant content. Then, when obtaining the class representation, weight is given to each support set sample, and the weighted class repre- sentation is used to better express the class. Moreover, a mutual similarity metric method is used to obtain a more accurate similarity relationship through the mutual similarity for each representation. Experiments prove that the approach in this paper performs well in few-shot image classification, and also shows remarkable excellence and competitiveness compared with related advanced techniques.
文献关键词：
中图分类号：
作者姓名：
ZHAO Wencang;QIN Wenqian;LI Ming
作者机构：
College of Automation and Electronic Engineering,Qingdao University of Science and Technology,Qingdao 266061,P.R.China
文献出处：
引用格式：
[1]ZHAO Wencang;QIN Wenqian;LI Ming-.Multi-attention fusion and weighted class representation for few-shot classification)[J].高技术通讯（英文版）,2022(03):295-306
A类：
MWNet
B类：
Multi,attention,fusion,weighted,representation,few,shot,classification,existing,learning,FSL,approaches,metric,usually,lack,distinction,feature,contributions,importance,each,sample,often,ignored,when,obtaining,where,performance,model,limited,Additional,similarity,method,also,worthy,Therefore,called,multi,WCR,proposed,this,paper,Firstly,module,introduced,into,highlight,valuable,part,reduce,interference,irrelevant,content,Then,given,support,set,used,better,express,Moreover,mutual,more,accurate,relationship,through,Experiments,prove,that,performs,well,image,shows,remarkable,excellence,competitiveness,compared,related,advanced,techniques
AB值：
0.491918
相似文献
机标中图分类号，由域田数据科技根据网络公开资料自动分析生成，仅供学习研究参考。
