语义匹配网络的小样本学习

doi:10.19678/j.issn.1000-3428.0056969

计算机工程 ›› 2021, Vol. 47 ›› Issue (5): 244-250,259. doi: 10.19678/j.issn.1000-3428.0056969

语义匹配网络的小样本学习

汪荣贵, 汤明空, 杨娟, 薛丽霞, 胡敏

合肥工业大学计算机与信息学院, 合肥 230601

收稿日期:2019-12-12 修回日期:2020-03-09 发布日期:2021-05-11
作者简介:汪荣贵(1966-),男,教授、博士生导师,主研方向为智能视频处理与分析、车载视觉增强系统;汤明空,硕士研究生;杨娟,博士;薛丽霞,副教授;胡敏,教授。
基金资助:
国家自然科学基金“基于视听信息融合的情感机器人情感识别与情感建模研究”（61672202）。

Semantic Matching Network for Few-Shot Learning

WANG Ronggui, TANG Mingkong, YANG Juan, XUE Lixia, HU Min

School of Computer Science and Information Engineering, Hefei University of Technology, Hefei 230601, China

Received:2019-12-12 Revised:2020-03-09 Published:2021-05-11

摘要/Abstract

摘要： 针对深度学习领域内通过少量样本难以实现视觉识别的小样本学习问题，提出一种新的语义匹配网络。利用双注意力机制匹配图像的语义信息，并在多尺度分类网络下匹配图像的相似度，提升同类别样本之间的语义相关性，从而获得更加准确的样本类别。实验结果表明，与Siamese Net、Matching Net等网络相比，该语义匹配网络可有效提取样本间的语义信息，提升小样本分类准确率。

关键词: 深度学习, 小样本学习, 语义匹配, 注意力机制, 特征提取

Abstract: In the field of deep learning,it is difficult to achieve visual recognition with a small number of samples.To address the problem,this paper proposes a semantic matching network.The dual attention mechanism is used to match the semantic information of the image,and the similarity of the image is matched under a multi-scale classification network to improve the semantic relevance between samples of the same category,so as to obtain more accurate sample categories.Experimental results show that the semantic matching network can effectively extract the semantic information between samples and improve the accuracy of few-shot classification.

Key words: deep learning, few-shot learning, semantic matching, attention mechanism, feature extraction

中图分类号:

TP391

汪荣贵, 汤明空, 杨娟, 薛丽霞, 胡敏. 语义匹配网络的小样本学习[J]. 计算机工程, 2021, 47(5): 244-250,259.

WANG Ronggui, TANG Mingkong, YANG Juan, XUE Lixia, HU Min. Semantic Matching Network for Few-Shot Learning[J]. Computer Engineering, 2021, 47(5): 244-250,259.

http://www.ecice06.com/CN/Y2021/V47/I5/244

参考文献

[1] KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[C]//Proceedings of NIPS'12.New York,USA:ACM Press,2012:84-90.
[2] SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[EB/OL].[2019-11-10].https://arxiv.org/abs/1409.4842.
[3] HUANG G,LIU Z,VAN DER M L,et al.Densely connected convolutional networks[EB/OL].[2019-11-10].https://arxiv.org/abs/1608.06993.
[4] REN S,HE K,GIRSHICK R B,et al.Faster R-CNN:towards real-time object detection with region proposal networks[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems.Washington D.C.,USA:IEEE Press,2015:91-99.
[5] REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2016:779-788.
[6] LIU W,ANGUEIOV D,ERHAN D,et al.SSD:single shot multibox detector[C]//Proceedings of European Conference on Computer Vision.Berlin,Germany:Springer,2016:21-37.
[7] ZHANG X T,QIANG Y T,SUNG F,et al.RelationNet2:deep comparison columns for few-shot learning[EB/OL].[2019-11-10].https://arxiv.org/abs/1811.07100.
[8] LI F F,FERGUS L,PERONA R.One-shot learning of object categories[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(4):594-611.
[9] LAKE B M,SALAKHUTDINOV R,GROSS J,et al.One shot learning of simple visual concepts[C]//Proceedings of Annual Meeting of the Cognitive Science Society.Boston,USA:[s.n.],2011:1-6.
[10] KOCH G,ZEMEL R,SALAKHUTDINOV R.Siamese neural networks for one-shot image recognition[EB/OL].[2019-11-10].https://www.cs.cmu.edu/~rsalakhu/papers/oneshot1.pdf.
[11] SNELL J,SWERSKY K,ZEMEL R S.Prototypical networks for few-shot learning[EB/OL].[2019-11-10].https://arxiv.org/abs/1703.05175.
[12] SUNG F,YANG Y,ZHANG W,et al.Learning to compare:relation network for few-shot learning[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2018:1199-1208.
[13] VINYALS O,BLUNDELL C,LILLICRAP T,et al.Matching networks for one shot learning[EB/OL].[2019-11-10].https://arxiv.org/abs/1606.04080.
[14] HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al.Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2016:770-778.
[15] VASWANI A,SHAZEER N,PARMAR N,et al.Attention is all you need[EB/OL].[2019-11-10].https://arxiv.org/abs/1706.03762.
[16] RAVI S,LAROCHELLE H.Optimization as a model for few-shot learning[C]//Proceedings of IEEE International Conference on Learning Representations.Washington D.C.,USA:IEEE Press,2017:157-168.
[17] FINN C,ABBEEL P,LEVINE S.Model-agnostic meta-learning for fast adaptation of deep networks[EB/OL].[2019-11-10].https://arxiv.org/abs/1703.03400.
[18] GARCIA V,BRUNA J.Few-shot learning with graph neural networks[EB/OL].[2019-11-10].https://arxiv.org/abs/1711.04043v1.
[19] GIDARIS S,KOMODAKIS N.Dynamic few-shot visual learning without forgetting[EB/OL].[2019-11-10].https://arxiv.org/abs/1804.09458.
[20] MUNKHDALIA T,YUAN X,MEHRI S,et al.Rapid adaptation with conditionally shifted neurons[EB/OL].[2019-11-10].https://arxiv.org/abs/1712.09926v3.
[21] QIAO Siyuan,LIU Chenxi,SHEN Wei,et al.Few-shot image recognition by predicting parameters from activations[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2018:7229-7238.
[22] HAN Zhidong.Dyna:a method of momentum for stochastic optimization[EB/OL].[2019-11-10].https://arxiv.org/abs/1805.04933.
[23] SHEN Hao.Towards a mathematical understanding of the difficulty in learning with feedforward neural networks[EB/OL].[2019-11-10].https://arxiv.org/abs/1611.05827.
[24] SELVARAJU R R,COGSWELL M,DAS A,et al.Grad-CAM:visual explanations from deep networks via gradient-based localization[J].International Journal of Computer Vision,2020,128(2):336-359.

选择文件类型/文献管理软件名称

选择包含的内容

语义匹配网络的小样本学习

Semantic Matching Network for Few-Shot Learning

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	丰芳宇, 罗晓曙, 蒙志明, 王广宇. 基于抗混叠残差注意力网络的人脸表情识别[J]. 计算机工程, 2023, 49(8): 190-198.
[2]	江雨燕, 陶承凤, 李平. 数据增强和自适应自步学习的深度子空间聚类算法[J]. 计算机工程, 2023, 49(8): 96-103, 110.
[3]	王书朋, 何引弟. 融合特征注意力机制的非均匀光照图像增强算法[J]. 计算机工程, 2023, 49(8): 232-239.
[4]	刘金硕, 王代辰, 邓娟, 王丽娜. 基于长尾分类算法的网络不良信息分类[J]. 计算机工程, 2023, 49(8): 13-19, 28.
[5]	刘昊鑫, 董超, 勾智楠, 高凯. 融合混合表征的小样本关系抽取方法[J]. 计算机工程, 2023, 49(8): 63-68.
[6]	李泽水, 冀俊忠, 杨翠翠. 基于边权重信息深度网络嵌入的PPIN功能模块检测[J]. 计算机工程, 2023, 49(8): 69-76.
[7]	杨长沛, 廖列法. 基于门控空洞卷积特征融合的中文命名实体识别[J]. 计算机工程, 2023, 49(8): 85-95.
[8]	王可铮, 徐玉芬, 周尚波. 结合对比感知损失和融合注意力的图像去雾模型[J]. 计算机工程, 2023, 49(8): 207-214.
[9]	刘俊豪, 王美林, 谢兴, 宋烨兴, 许莉花. 基于改进YOLOv5的皮革瑕疵检测算法[J]. 计算机工程, 2023, 49(8): 240-249.
[10]	马娜, 温廷新, 贾旭, 李晓会. 复杂光照条件下自适应的车脸重识别模型[J]. 计算机工程, 2023, 49(8): 275-282, 290.
[11]	陈露萌, 曹彦彦, 黄民, 谢鑫钢. 基于改进YOLOv5的火焰检测方法[J]. 计算机工程, 2023, 49(8): 291-301, 309.
[12]	李强龙, 周新文, 位梦恩, 甘阳洲. 基于条形池化和注意力机制的街道场景红外目标检测算法[J]. 计算机工程, 2023, 49(8): 310-320.
[13]	张家熔, 苑津莎, 许珈宁, 罗志宏. 基于多元信息嵌入与协同神经网络的力学实体识别算法[J]. 计算机工程, 2023, 49(7): 125-134.
[14]	白明昌. 基于折叠路径聚合的属性网络节点嵌入方法[J]. 计算机工程, 2023, 49(7): 76-84.
[15]	闫兴亚, 匡娅茜, 白光睿, 李月. 基于深度学习的学生课堂行为识别方法[J]. 计算机工程, 2023, 49(7): 251-258.

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

语义匹配网络的小样本学习

Semantic Matching Network for Few-Shot Learning

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价