基于目标域局部近邻几何信息的域自适应图像分类方法

doi:10.11772/j.issn.1001-9081.2017.04.1164

计算机应用 ›› 2017, Vol. 37 ›› Issue (4): 1164-1168.DOI: 10.11772/j.issn.1001-9081.2017.04.1164

• 计算机视觉与虚拟现实 • 上一篇下一篇

基于目标域局部近邻几何信息的域自适应图像分类方法

唐宋^1,2, 陈利娟¹, 陈志贤³, 叶茂¹

1. 电子科技大学计算机科学与工程学院, 成都 611731;
2. 重庆邮电大学复杂系统分析与控制研究中心, 重庆 400065;
3. 中国科学院深圳先进技术研究院, 广东深圳 518055

收稿日期:2016-09-05 修回日期:2016-12-21 出版日期:2017-04-10 发布日期:2017-04-19
通讯作者: 陈利娟
作者简介:唐宋(1982-),男,四川成都人,博士研究生,主要研究方向:机器学习、神经网络;陈利娟(1986-),女,山西大同人,博士研究生,主要研究方向:机器学习、动力系统稳定性;陈志贤(1988-),男,河北井陉人,博士研究生,主要研究方向:机器人智能控制;叶茂(1972-),男,重庆人,教授,博士,主要研究方向:模式识别、神经网络。
基金资助:
国家自然科学基金资助项目（61375038，61501073）；重庆市杰出青年基金资助项目（cstc2013jcyjjq40001）；四川省应用基础研究计划项目（2016JY0088）。

Domain adaptation image classification based on target local-neighbor geometrical information

TANG Song^1,2, CHEN Lijuan¹, CHEN Zhixian³, YE Mao¹

1. School of Computer Science & Engineering, University of Electronic Science and Technology of China, Chengdu Sichuan 611731, China;
2. Center of Analysis and Control for Complex Systems, Chongqing University of Posts and Telecommunications, Chongqing 400065, China;
3. Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen Guangdong 518055, China

Received:2016-09-05 Revised:2016-12-21 Online:2017-04-10 Published:2017-04-19
Supported by:
This work is partially supported by the National Natural Science Foundation of China (61375038,61501073), the Science Fund for Distinguished Young Scholars of Chongqing (cstc2013jcyjjq40001), the Applied Basic Research Programs of Sichuan Science and Technology Department (2016JY0088).

摘要/Abstract

摘要： 在许多实际工程应用中，训练场景（源域）和测试场景（目标域）的分布并不相同，如果将源域中训练的分类器直接应用到目标域，性能往往会出现大幅度下降。目前大多数域自适应方法以概率推导为基础。从图像特征表达的角度出发，针对自适应图像分类问题，提出一种新的基于协同特征的无监督方法。首先，所有源样本被作为字典；然后，距离目标样本最近的三个目标域样本被用来帮助鲁棒地表达局部近邻几何信息；最后，结合字典和局部近邻信息实现编码，并利用最近邻分类器完成分类。因为协同特征通过融合目标域局部近邻信息，获得了更强的鲁棒性和区分性，基于该特征编码的分类方法具有更好的分类性能。在域自适应数据集上的对比实验结果表明所提算法是有效的。

关键词: 域自适应, 流形, 目标域局部近邻关系, 协同表达, 图像分类

Abstract: In many real engineering applications, the distribution of training scenarios (source domain) and the distribution of testing scenarios (target domain) is different, thus the classification performance decreases sharply when simply applying the classifier trained in source domain directly to the target domain. At present, most of the existing domain adaptation methods are based on the probability-inference. For the problem of domain adaptation image classification, a collaborative representation based unsupervised method was proposed from the view of image representation. Firstly, all of the source samples were taken as the dictionary. Secondly, the three target samples closest to the target sample in the target domain were exploited to robustly represent the local-neighbor geometrical information. Thirdly, the target sample was encoded by combining the dictionary and the local-neighbor information. Finally, the classification was completed by using the nearest classifier. Since the collaborative representations have stronger robustness and discriminative ability by absorbing the target local-neighbor information, the classification method based on the new representations has better classification performance. The experimental results on the domain adaptation dataset confirm the effectiveness of the proposed method.

Key words: domain adaptation, manifold, target local-neighbor information, collaborative representation, image classification

中图分类号:

TP181

唐宋, 陈利娟, 陈志贤, 叶茂. 基于目标域局部近邻几何信息的域自适应图像分类方法[J]. 计算机应用, 2017, 37(4): 1164-1168.

TANG Song, CHEN Lijuan, CHEN Zhixian, YE Mao. Domain adaptation image classification based on target local-neighbor geometrical information[J]. Journal of Computer Applications, 2017, 37(4): 1164-1168.

参考文献

[1] 刘建伟, 孙正康, 罗雄麟. 域自适应学习研究进展[J]. 自动化学报, 2014, 40(8):1576-1600.(LIU J W, SUN Z K, LUO X L. Review and research development on domain adaptation learning[J]. Acta Automatica Sinica, 2014, 40(8):1576-1600.)
[2] QIU Q, PATEL V, TURAGA R, et al. Domain adaptive dictionary learning[C]//ECCV 2012: Proceedings of 13st European Conference on Computer Vision. Berlin: Springer, 2012: 631-645.
[3] CHEN M, XU Z, WEINBERGER K, et al. Marginalized denoising autoencoders for domain adaptation[C]//ICML 2012: Proceedings of 29th International Conference on Machine Learning. New York: ACM, 2012:115-122.
[4] GONG B, GRAUMAN K, SHA F. Connecting the dots with landmarks: discriminatively learning domain-invariant features for unsupervised domain adaptation[C]//ICML 2013: Proceedings of 30th International Conference on Machine Learning. New York: ACM, 2013: 222-230.
[5] GOPALAN R, LI R, CHELLAPPA R. Domain adaptation for object recognition: an unsupervised approach[C]//ICCV 2011: Proceedings of 13rd International Conference on Computer Vision. Piscataway, NJ: IEEE, 2011: 999-1006.
[6] GONG B, SHI Y, SHA F, et al. Geodesic flow kernel for unsupervised domain adaptation[C]//CVPR 2012: Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition. Washington, DC: IEEE Computer Society, 2012: 2066-2073.
[7] WAQAS J, ZHANG Y, ZHANG L. Collaborative neighbor representation based classification using l2-minimization approach[J]. Pattern Recognition Letters, 2013, 34(2): 201-208.
[8] TANG S, YE M, LIU Q, et al. Domain adaptation of image classification based on collective target nearest-neighbor representation[J]. Journal of Electronic Imaging, 2016, 25(3): 033006-033015.
[9] SAENKO K, KULIS B, FRITZ M, et al. Adapting visual category models to new domains[C]//ECCV 2010: Proceedings of the European Conference on Computer Vision. Berlin: Springer, 2010: 213-226.
[10] NI J, QIU Q, CHELLAPPA R. Subspace interpolation via dictionary learning for unsupervised domain adaptation[C]//CVPR 2013: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, DC: IEEE Computer Society, 2013: 692-699.
[11] GONG B, GRAUMAN K, SHA F. Reshaping visual datasets for domain adaptation[EB/OL].[2016-05-10].http://www.cs.utexas.edu/~grauman/papers/gong-NIPS2013.pdf.
[12] CUI Z, LI W, XU D, et al. Flowing on Riemannian manifold: domain adaptation by shifting covariance[J]. IEEE Transactions on Cybernetics, 2014, 44(12): 2264-2273.
[13] SI S, TAO D, GENG B, BREGMAN. Divergence-based regularization for transfer subspace learning[J]. IEEE Transactions on Knowledge Data Engineering, 2010, 22(7): 929-942.
[14] PAN S J, TSANG I W, KWOK J T, et al. Domain adaptation via transfer component analysis[J]. IEEE Transaction on Neural Network, 2011, 22(2): 199-210.

[1]	贾鹤鸣, 郎春博, 姜子超. 基于轻量级卷积神经网络的植物叶片病害识别方法[J]. 计算机应用, 2021, 41(6): 1812-1819.
[2]	刘晓龙, 王士同. 渐进式分离的开放集模糊域自适应算法[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3127-3131.
[3]	许学斌, 张佳达, 刘伟, 路龙宾, 赵雨晴. 融合空间和通道特征的高精度乳腺癌分类方法[J]. 计算机应用, 2021, 41(10): 3025-3032.
[4]	郑延斌, 韩梦云, 樊文鑫. 基于二维主成分分析与卷积神经网络的手写体汉字识别[J]. 计算机应用, 2020, 40(8): 2465-2471.
[5]	边小勇, 江沛龄, 赵敏, 丁胜, 张晓龙. 基于多分支神经网络模型的弱监督细粒度图像分类方法[J]. 计算机应用, 2020, 40(5): 1295-1300.
[6]	吴立人, 刘政浩, 张浩, 岑悦亮, 周维. 聚焦图像对抗攻击算法PS-MIFGSM[J]. 计算机应用, 2020, 40(5): 1348-1353.
[7]	徐戈, 肖永强, 汪涛, 陈开志, 廖祥文, 吴运兵. 基于视觉误差与语义属性的零样本图像分类[J]. 计算机应用, 2020, 40(4): 1016-1022.
[8]	郭志强, 胡永武, 刘鹏, 杨杰. 基于特征融合的室外天气图像分类[J]. 计算机应用, 2020, 40(4): 1023-1029.
[9]	张凯琳, 阎庆, 夏懿, 章军, 丁云. 基于焦点损失的半监督高光谱图像分类[J]. 计算机应用, 2020, 40(4): 1030-1037.
[10]	张盼盼, 李其申, 杨词慧. 基于轻量级分组注意力模块的图像分类算法[J]. 计算机应用, 2020, 40(3): 645-650.
[11]	史梦安, 陆振宇. 基于YOLO-tiny-RFB模型的电站旋钮开关状态识别[J]. 计算机应用, 2020, 40(12): 3679-3686.
[12]	邓棋, 雷印杰, 田锋. 用于肺炎图像分类的优化卷积神经网络方法[J]. 计算机应用, 2020, 40(1): 71-76.
[13]	武宽, 秦品乐, 柴锐, 曾建朝. 基于不同超声成像的甲状腺结节良恶性判别[J]. 计算机应用, 2020, 40(1): 77-82.
[14]	何新宇, 张晓龙. 基于深度神经网络的肺炎图像识别模型[J]. 计算机应用, 2019, 39(6): 1680-1684.
[15]	杨洋, 王正群, 徐春林, 严陈, 鞠玲. 基于最大间隔准则的鲁棒多流形判别局部图嵌入算法[J]. 计算机应用, 2019, 39(5): 1453-1458.

基于目标域局部近邻几何信息的域自适应图像分类方法

Domain adaptation image classification based on target local-neighbor geometrical information

PDF

可视化

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