基于码字匹配和引力筛选的半监督协同训练算法

陈善学; 尹修玄; 杨亚娟

doi:10.13203/j.whugis20130840

基于码字匹配和引力筛选的半监督协同训练算法

重庆邮电大学移动通信技术重庆市重点实验室, 重庆, 400065

基金项目: 国家自然科学基金资助项目(61271260,61102062);重庆市教委科学技术研究资助项目(KJ1400416)。

详细信息

作者简介:
陈善学,博士,教授,主要从事图像处理、数据压缩方面的研究。E-mail:chee420@163.com

通讯作者:
尹修玄,硕士生。E-mail:492989162@qq.com

中图分类号: P237.3
计量
- 文章访问数: 1721
- HTML全文浏览量: 63
- PDF下载量: 302
出版历程
- 收稿日期: 2014-07-09
- 发布日期: 2015-10-04

Semi-Supervised Collaboration Training Algorithm Based on Codeword Matching and Gravitation Selecting

Chongqing Key Laboratory of Mobile Communications Technology, Chongqing University of Posts and Telecommunications, Chongqing 400065, China

Funds: The National Natural Science Foundation of China, Nos.61271260, 61102062; the Scientific and Technological Research Program of Chongqing Municipal Education Commission, No. KJ1400416.

摘要

摘要: 针对传统的Co-training和Tri-training协同训练算法中基分类器独立性低、迭代过程中误差累积和整体泛化性能低的问题,将多视图理论、编码理论和万有引力公式引入协同训练分类算法中,提出了改进算法,算法有效地防止了迭代过程中的误差累积,同时提高了分类系统的泛化性能。在高光谱图像分类实验中,随机地从数据集中抽取5%、10%和20%样本作为已标记训练集时,码字匹配的协同训练分类算法对比Co-training和Tri-training算法,在分类精度上平均分别提高了12.38%和6.13%,在Kappa系数上平均分别提高了0.2和0.07。进一步加入引力筛选机制,对比Co-training和Tri-training算法,在分类精度上平均分别提高了21.30%和10.99%,在Kappa系数上平均分别提高了0.26和0.13,结果表明了本文算法的有效性。
- 高光谱图像分类 /
- 协同训练 /
- 码字匹配 /
- 万有引力
Abstract: Traditional collaboration training algorithms, such as Co-training and Tri-training, has some problems: low independence of the base classifier, error accumulation during iteration and low generalization system performance. Accordingly, this paper proposes an import multi-view theory,coding theory and formula for universal gravitation as applied to the collaboration training algorithm preventing both error accumulation and improving generalization performance at the same time. During a hyper-spectral image classification experiment, randomly selecting 5%,10% and 20% samples from data sets as a labled train set, the collaboration training algorithm for codeword matching had a 12.38% and 6.13% higher accuracy matching rate than Co-training and Tri-training respectively. At the same time, it had a respective 0.2 and 0.07 higher Kappa coefficient. In contrast, a collaboration training algorithm for classification based on codeword matching and gravitation selecting had 21.30% and 10.99% higher accuracy than Co-training and Tri-training and 0.26 and 0.17 higher Kappa coefficient. These results demonstrate the validity of the proposed algorithm.
- hyperspectral image classification /
- collaboration training /
- code matching /
- universal gravitation

HTML全文

参考文献(21)

[1]	Wang Y, Chen S, Zhou Z H. New Semi-supervised Classification Method Based on Modified Cluster Assumption[J]. IEEE Transactions on Neural Networks and Learning Systems, 2012, 23(5): 689-702
[2]	Shen W, Wu G, Sun Z, et al. Study on Classification Methods of Remote Sensing Image Based on Decision Tree Technology[C]. IEEE International Conference on Computer Science and Service System (CSSS), Nanjing,China,2011
[3]	Niu D, Wang F, Chang Y, et al. An Improved Multi-objective Differential Evolution Algorithm[C]. The 24th IEEE Chinese Control and Decision Conference (CCDC), Taiyuan,China,2012
[4]	Yan W, Yang W, Sun H, et al. Unsupervised Classification of PolInSAR Data Based on Shannon Entropy Characterization with Iterative Optimization[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2011, 4(4): 949-959
[5]	Wang Y, Chen S. Safety-Aware Semi-Supervised Classification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2013, 24(11): 1 763-1 772
[6]	Xiong Biao, Jiang Wanshou, Li Lelin. Gauss Mixture Model Based Semi-Supervised Classification for Remote Sensing Image [J].Geomatics and Information Science of Wuhan University,2011, 36(1): 108-112(熊彪,江万寿,李乐林.基于高斯混合模型的遥感影像半监督分类[J]. 武汉大学学报·信息科学版,2011, 36(1): 108-112)
[7]	Yan Li, Nie Qian, Hu Wenyuan,et al. Rapidly Computing Satellite Gravity Gradient Observations Along Orbit from Gravity Field Model [J].Geomatics and Information Science of Wuhan University, 2009, 34(2): 183-186(闫利,聂倩,胡文元,等.基于对象级的ADS40遥感影像分类研究[J]. 武汉大学学报·信息科学版,2009, 34(2): 183-186)
[8]	Wang Y, Chen S, Zhou Z H. New Semi-Supervised Classification Method Based on Modified Cluster Assumption[J]. IEEE Transactions on Neural Networks and Learning Systems, 2012, 23(5): 689-702
[9]	Xiong B, Jiang W, Zhang F. Semi-Supervised Classification Considering Space and Spectrum Constraint for Remote Sensing Imagery[C]. The 18th IEEE International Conference on Geoinformatics, Beijing,China,2010
[10]	Dopido I, Li J, Plaza A, et al. Semi-Supervised Classification of Hyperspectral Data Using Spectral Unmixing Concepts[C]. IEEE Tyrrhenian Workshop on Advances in Radar and Remote Sensing (TyWRRS), Naples, Italia,2012
[11]	Yan J, Yun X, Wu Z, et al. Online Traffic Classification Based on Co-training Method[C].The 13th International Conference on Parallel and Distributed Computing, Applications and Technologies, IEEE Computer Society, Beijing,China,2012
[12]	Du W, Phlypo R, Adali T. Adaptive Feature Split Selection for Co-training: Application to Tire Irregular Wear Classification[C]. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Vancouver, BC,2013
[13]	Nigam K, Ghani R. Analyzing the Effectiveness and Applicability of Co-training[C].The 9th International Conference on Information and Knowledge Management, McLean, VA, 2000
[14]	Xu J, He H, Man H. DCPE Co-training: Co-training Based on Diversity of Class Probability Estimation[C]. The IEEE International Joint Conference on Neural Networks (IJCNN), Barcelona,Spain,2010
[15]	Zhou Z H, Li M. Tri-training: Exploiting Unlabeled Data Using Three Classifiers[J]. IEEE Transactions on Knowledge and Data Engineering, 2005, 17(11): 1 529-1 541
[16]	Du Bo, Zhang Liangpei, Li Pingxiang, et al. A Constrained Energy Minimization Method in Sub-pixel Target Detection Based on Minimization Noise Fraction [J].Journal of Image and Graphics, 2009, 14 (9):1 850-1 857(杜博,张良培,李平湘,等.基于最小噪声分离的约束能量最小化亚像元目标探测方法[J].中国图像图形学报,2009, 14 (9):1 850-1 857)
[17]	Huang R, He W. Using Tri-training to Exploit Spectral and Spatial Information for Hyperspectral Data Classification[C]. IEEE International Conference on Computer Vision in Remote Sensing (CVRS), Xiamen,China,2012
[18]	Zhang Yan, Lv Danju, Wu Baoguo. Research of Semi-Supervised Classification Algorithm Based on Tri-training [J].Computer Technology and Development, 2013, 23(7): 77-79(张雁, 吕丹桔, 吴保国. 基于 Tri-Training 半监督分类算法的研究[J].计算机技术与发展, 2013, 23(7): 77-79)
[19]	Liu X, Zhang H, Cai Z, et al. A Tri-training Based Transfer Learning Algorithm[C].The 24th IEEE International Conference on Tools with Artificial Intelligence (ICTAI), Athens,Greece,2012
[20]	Bender T, Kjaer T W, Thomsen C E, et al. Semi-Supervised Adaptation in Ssvep-based Brain-computer Interface Using Tri-training[C].The 35th IEEE Annual International Conference of Engineering in Medicine and Biology Society (EMBC), Osaka,Japan,2013
[21]	Okabe M, Yamada S. Clustering with Extended Constraints by Co-training[C]. IEEE/WIC/ACM International Conferences on Web Intelligence and Intelligent Agent Technology (WI-IAT), Macau,China,2012

施引文献(6)

期刊类型引用(2)

1.	吴疆，岳贤亮，董婷，蒋平. 单分类支持向量机用于样本不平衡数据集建模研究. 微型电脑应用. 2020(06): 81-82+92 . 百度学术
2.	李恋，柴豪森，徐浩. 基于K-means算法的Co-trainning的研究. 电脑知识与技术. 2020(32): 216-217 . 百度学术