[1]孙乐,王士同.不确定成对约束的双对抗流形传播方法[J].智能系统学报,2023,18(2):270-281.[doi:10.11992/tis.202202025]
SUN Le,WANG Shitong.Doubly adversarial manifold propagation on uncertain pairwise constraints[J].CAAI Transactions on Intelligent Systems,2023,18(2):270-281.[doi:10.11992/tis.202202025]
点击复制
SUN Le,WANG Shitong.Doubly adversarial manifold propagation on uncertain pairwise constraints[J].CAAI Transactions on Intelligent Systems,2023,18(2):270-281.[doi:10.11992/tis.202202025]
不确定成对约束的双对抗流形传播方法
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
18
期数:
2023年第2期
页码:
270-281
栏目:
学术论文—机器学习
出版日期:
2023-05-05
- Title:
- Doubly adversarial manifold propagation on uncertain pairwise constraints
- Keywords:
- uncertain pairwise constraint; doubly adversarial relationship; manifold regularization; pairwise constraint propagation; possibility of constraints; k-nearest neighbor; distinguishability; weak supervision
- 分类号:
- TP181
- 摘要:
- 成对约束传播(pairwise constraint propagation, PCP)通常研究的是在初始给定精确的成对约束基础上通过传播学习来增加成对约束的数量，从而给机器学习任务提供较多的监督信息。可是，在现实场景中，有时还有一些不精确的成对约束，因此，如何利用这些不精确的成对约束来提高成对约束传播学习的效果是一个有待解决的问题。针对这一问题，本文提出了一种不确定成对约束的传播方法。主要思想是用两个矩阵分别表示必须链接和不能链接的可能性，两种可能性之间形成对抗，同时两种成对约束之间也存在对抗关系，两类对抗相结合形成一种双对抗结构，作用于必须链接和不能链接的传播过程，使二者的对抗强度在竞争中趋于最小化。我们将该方法称为不确定成对约束传播(uncertain pairwise constraint propagation, UPCP)。在多个数据集上的实验结果表明，不确定成对约束的传播效果不超过但近似于理想化传播效果，在增强现实应用性的同时尽可能地保证了传播精度。
- Abstract:
- Pairwise constraint propagation aims to increase the number of accurate pairwise constraints through propagation learning on the basis of the initial given exact pairwise constraints to provide additional supervision information for the machine learning task. However, inaccurate pairwise constraints occasionally exist in real scenes. Therefore, one problem to be solved is the use of these inaccurate pairwise constraints to enhance pairwise constraint propagation. A propagation method of uncertain pairwise constraints is proposed in this paper to solve this problem. The main idea is to use two matrices to represent the possibility of must-links and cannot-links. A confrontation between the two possibilities and a confrontation relationship between two kinds of pairwise constraints exist. The combination of two types of confrontation forms a double adversarial structure, which acts on the propagation process of must-links and cannot-links; therefore, the adversarial intensity between them is minimized in the competition. This method is called uncertain pairwise constraint propagation. Experimental results on multiple datasets show that the propagation effect of uncertain pairwise constraints does not exceed but is similar to the ideal propagation effect, which enhances the practical applicability and ensures the propagation accuracy completely.
- 参考文献/References:
-
[1] FU Z Y, LU Z W. Pairwise constraint propagation: a survey[J]. Computer science, 2015, 1(3): 161–164.
[2] 付振勇. 成对约束传递方法及其在约束聚类问题中的应用研究[D]. 上海: 上海交通大学, 2012.
FU Zhenyong. Studies on pairwise constraint propagation and its applications in the constrained clustering tasks[D]. Shanghai: Shanghai Jiao Tong University, 2012.
[3] WAGASTAFF K, CARDIE C, ROGERS S, et al. Constrained k-means clustering with background knowledge[C]// Eighteenth International Conference on Machine Learning. San Francisco: Morgan Kaufmann Publishers Inc, 2001: 577?584.
[4] KLEIN D. From instance-level constraints to space-level constraints: Making the most of prior knowledge in data clustering[J]. Proceedings of the Nineteenth International Conference on Machine Learning, 2002, 12(7): 307–314.
[5] BASU S, BILENKO M, MOONEY R J. A probabilistic framework for semi-supervised clustering[C]// Proceedings of the 10th Association for Computing Machinery (ACM) Special Interest Group on Knowledge Discovery and Data Mining (SIGKDD) International Conference on Knowledge Discovery and Data Mining. Seattle: Dept. of Computer Sciences University of Texas at Austin, 2004: 59?68.
[6] KULIS B, BASU S, DHILLON I, et al. Semi-supervised graph clustering: a kernel approach[J]. Machine learning, 2009, 74(1): 1–22.
[7] 卞则康, 王士同. 基于混合距离学习的鲁棒的模糊C均值聚类算法[J]. 智能系统学报, 2017, 12(4): 450–458
BIAN Zekang, WANG Shitong. Robust FCM clustering algorithm based on hybrid-distance learning[J]. CAAI transactions on intelligent systems, 2017, 12(4): 450–458
[8] WANG D, GAO X B, WANG X M. Semi-supervised nonnegative matrix factorization via constraint propagation[J]. IEEE transactions on cybernetics, 2016, 46(1): 233–244.
[9] CHEN S G, ZHANG D Q. Semi-supervised dimensionality reduction with pairwise constraints for hyperspectral image classification[J]. IEEE geoscience and remote sensing letters, 2011, 8(2): 369–373.
[10] 钱进,罗鼎. 基于成对约束半监督降维的高光谱遥感影像特征提取[J]. 遥感技术与应用, 2014, 29(4): 681–688
QIAN Jin, LUO Ding. Feature extraction from hyperspectral remote sensing imagery based on semisupervised dimensionality reduction with pairwise constraints[J]. Remote sensing technology and application, 2014, 29(4): 681–688
[11] LIU Mingxia, ZHANG Daoqiang. Pairwise constraint-guided sparse learning for feature selection[J]. IEEE transactions on cybernetics, 2017, 46(2): 298–310.
[12] 赵帅, 张雁, 徐海峰. 基于成对约束分的特征选择及稳定性评价[J]. 计算机与数字工程, 2019, 47(6): 1441–1445
ZHAO Shuai, ZHANG Yan, XU Haifeng. Feature selection and stability evaluation based on pairwise constraint score[J]. Computer & digital engineering, 2019, 47(6): 1441–1445
[13] XING E P, NG A Y, JORDAN M I, et al. Distance metric learning with application to clustering with side-information[C]// Advances in Neural Information Processing Systems. Cambridge, Boston: Massachusetts Institute of Technology Press, 2002: 318?326.
[14] HOI S C H, LIU Wei, LYU M R, et al. Learning distance metrics with contextual constraints for image retrieval[C]//2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. New York, USA: 2072?2078.
[15] LIU Wei, MA Shiqian, TAO Dacheng, et al. Semi-supervised sparse metric learning using alternating linearization optimization[C]// Acm Sigkdd International Conference on Knowledge Discovery & Data Mining. New York: ACM, 2010: 1139?1148.
[16] FU Zhenyong, LU Zhiwu, IP H H S, et al. Local similarity learning for pairwise constraint propagation[J]. Multimedia tools and applications, 2015, 74(11): 3739–3758.
[17] 刘笑, 陈家炜, 胡峻林. 用于亲属关系鉴别的成对约束组合度量学习[J]. 山东大学学报(工学版), 2022, 52(2): 50–56
LIU Xiao, CHEN Jiawei, HU Junlin. Pairwise constrained compositional metric learning for kinship verification[J]. Journal of Shandong University (engineering science edition), 2022, 52(2): 50–56
[18] YAN Rong, ZHANG Jian, YANG Jie, et al. A discriminative learning framework with pairwise constraints for video object classification[J]. IEEE transactions on pattern analysis and machine intelligence, 2006, 28(4): 578–593.
[19] LU Zhengdong, CARREIRA-PERPINAN M A. Constrained spectral clustering through affinity propagation[C]//IEEE Conference on Computer Vision and Pattern Recognition. Anchorage: IEEE, 2008: 1?8.
[20] LI Z, LIU J, TANG X. Pairwise constraint propagation by semidefinite programming for semi-supervised classification[C]//Machine Learning, Proceedings of the Twenty-Fifth International Conference. Helsinki, Finland, 2008: 264?272.
[21] YU S X, SHI J B. Segmentation given partial grouping constraints[J]. IEEE transactions on pattern analysis and machine intelligence, 2004, 26(2): 173–183.
[22] LU Z, PENG Y. Exhaustive and efficient constraint propagation: a graph-based learning approach and its applications[J]. International journal of computer vision, 2013, 103(3): 306–325.
[23] JIA Y, LIU H, HOU J. Pairwise constraint propagation with dual adversarial manifold regularization[J]. IEEE transactions on neural networks and learning systems., 2020, 31(12): 5575–5587.
[24] LU Z, PENG Y. Heterogeneous constraint propagation with constrained sparse representation[C]// IEEE International Conference on Data Mining. Dallas: IEEE, 2013.
[25] FU Z, IP H S, LU H, et al. Multi-modal constraint propagation for heterogeneous image clustering[C] //Proceedings of the 19th ACM international conference on Multimedia. Scottsdale: ACM, 2011: 143–152.
[26] BERAHMAND K, ROSTOMI M, FOROUZANDEH S. A novel method of constrained feature selection by the measurement of pairwise constraints uncertainty[J]. Journal of big data, 2020, 83(7): 483–504.
[27] 郑静, 熊伟丽. 基于互信息的多块k近邻故障监测及诊断[J]. 智能系统学报, 2021, 16(4): 717–728
ZHENG Jing, XIONG Weili. Multiblock k-nearest neighbor fault monitoring and diagnosis based on mutual information[J]. CAAI transactions on intelligent systems, 2021, 16(4): 717–728
[28] LEE D D, SEUNG H S. Algorithms for non-negative matrix factorization[C]// International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2000: 535–541.
[29] 卞则康, 张进, 王士同. 基于类标感知的KNN分类算法[J]. 模式识别与人工智能, 2021, 34(10): 873–884
BIAN Zekang, ZHANG Jin, WANG Shitong. Class-aware based KNN classification method[J]. Pattern recognition and artificial intelligence, 2021, 34(10): 873–884
[30] FU Z Y, LU Z W, IP H, et al. Symmetric graph regularized constraint propagation[J]. Proceedings of the AAAI conference on artificial intelligence, 2011, 25(1): 350–355.
备注/Memo
收稿日期:2022-02-26。
基金项目:国家自然科学基金项目（61972181）.
作者简介:孙乐,硕士研究生,主要研究方向为人工智能、模式识别;王士同,教授,博士生导师,主要研究方向为人工智能与模式识别。发表学术论文近百篇
通讯作者:王士同. E-mail：wxwangst@aliyun.com
更新日期/Last Update:
1900-01-01