一种基于类别不平衡数据的层次分类模型

doi:10.3969/j.issn.0253-2778.2015.01.010

中国科学技术大学学报 ›› 2015, Vol. 45 ›› Issue (1): 61-68.DOI: 10.3969/j.issn.0253-2778.2015.01.010

• 论著 • 上一篇

一种基于类别不平衡数据的层次分类模型

施培蓓，刘贵全，汪中，卫兵

1.合肥师范学院公共计算机教学部，安徽合肥 230601；2.中国科学技术大学计算机学院，安徽合肥 230027； 3.中国电子科技集团公司第三十八研究所数字技术部,安徽合肥 230088

收稿日期:2014-06-09 修回日期:2014-07-29 接受日期:2014-07-29 出版日期:2014-07-29 发布日期:2014-07-29
通讯作者: 施培蓓
作者简介:施培蓓(通讯作者),女,1984年生，硕士/讲师. 研究方向：数据挖掘、机器学习. E-mail: pb_shi@163.com
基金资助:
国家科技支撑计划（2012BAH17B03），安徽省自然科学基金（1408085MF131），安徽省高等学校自然科学项目（KJ2013B212），合肥师范学院魂芯DSP产业化研究院开放课题资助.

A hierarchical classification model for class-imbalanced data

SHI Peibei, LIU Guiquan, WANG Zhong, WEI Bing

1.Department of Public Computer Teaching, Hefei Normal University, Hefei 230601, China; 2.School of Computer Science and Technology, University of Science and Technology of China，Hefei 230027, China; 3.Department of Digital Technology, No.38 Research Institute of CETC, Hefei 230088, China

Received:2014-06-09 Revised:2014-07-29 Accepted:2014-07-29 Online:2014-07-29 Published:2014-07-29

摘要/Abstract

摘要： 传统的机器学习方法在处理类别不平衡数据时分类性能较低，为此提出一种基于类别不平衡数据的层次分类模型.层次分类模型采用AdaBoost方法为基准分类器，以分类器误报率和特征建立数学模型，并证明层次分类模型的参数可以计算得到.首先以层次分类树为结构建立模型，接着针对层次分类树的结构模型进行分类代价计算，得到模型的代价与每层特征之间的定量数学描述，然后将该分类代价转换为优化问题并给出优化问题的求解过程，同时给出层次分类模型的计算结果.在UCI数据集上进行大量测试，以AUC和F-Measure为评价标准，相比于现有的不平衡分类方法，层次分类模型具有更优的分类性能.

关键词: 机器学习, 类别不平衡, 层次分类, 特征, 评价标准

Abstract: Traditional machine learning methods have lower classification performance when dealing with class imbalanced data. A hierarchical classification model for class imbalanced data was thus proposed. With an AdaBoost classifier as its basis classifier, the model builds mathematical models by the features and false positive rates of the classifier, and demonstrates that parameters of the hierarchical classification model could be calculated. First, the hierarchical classification tree was as the structure, and then the classification cost of the hierarchical classification tree mode was obtained as well as a quantitative and mathematical description of the features of each layer. Finally, the classification cost could be converted to a optimization problem, and the solving process of the optimization problem was given. Meanwhile, results of the hierarchical classification are presented. Experiments have been conducted on UCI dataset, and the results show that the proposed method has higher AUC and F-measure compared to many existing class-imbalanced learning methods.

Key words: machine learning, class-imbalanced, hierarchical classification, feature, evaluation criteria

中图分类号:

TP391

施培蓓，刘贵全，汪中，卫兵. 一种基于类别不平衡数据的层次分类模型[J]. 中国科学技术大学学报, 2015, 45(1): 61-68.

SHI Peibei, LIU Guiquan, WANG Zhong, WEI Bing. A hierarchical classification model for class-imbalanced data[J]. Journal of University of Science and Technology of China, 2015, 45(1): 61-68.

参考文献

［1］ Phua C, Alahakoon D, Lee V. Minority report in fraud detection: classification of skewed data［J］. ACM SIGKDD Explorations Newsletter, 2004, 6(1): 50-59.
［2］ Sun A X, Lim E P, Liu Y. On strategies for imbalanced text classification using SVM: A comparative study［J］. Decision Support Systems, 2009, 48(1): 191-201.
［3］ Turney P D. Learning algorithms for key phrase extraction［J］. Information Retrieval, 2000, 2(4): 303-336.
［4］ Burez J, van den Poel D. Handling class imbalance in customer churn prediction［J］. Expert Systems with Applications, 2009, 36(3): 4 626-4 636.
［5］ Brekke C, Solberg A H S. Oil spill detection by satellite remote sensing［J］. Remote sensing of environment, 2005, 95(1): 1-13.
［6］ Plant C, Bhm C, Tilg B, et al. Enhancing instance-based classification with local density: a new algorithm for classifying unbalanced biomedical data［J］. Bioinformatics, 2006, 22(8): 981-988.
［7］ Branch J W, Giannella C, Szymanski B, et al. In-network outlier detection in wireless sensor networks［J］. Knowledge and information systems, 2013, 34(1): 23-54.
［8］ Sahbi H, Geman D. A hierarchy of support vector machines for pattern detection［J］. Journal of Machine Learning Research, 2006, 7: 2 087-2 123.
［9］ Blake C, Keogh E, Merz C J. UCI repository of machine learning databases［EB/OL］. http://www.ics.uci.edu/_mlearn/MLRepository.html.
［10］ Chawla N V, Bowyer K W, Hall L O, et al. SMOTE: Synthetic minority over-sampling technique［J］. Journal of Artificial Intelligence Research, 2002, 16: 321-357.
［11］ Han H, Wang W Y, Mao B H. Borderline-SMOTE: A new over-sampling method in imbalanced data sets learning［C］//Advances in Intelligent Computing. Berlin Heidelberg, Germany: Springer, 2005: 878-887.
［12］ Liu A, Ghosh J, Martin C E. Generative oversampling for mining imbalanced datasets［C］// Proceedings of International Conference on Data Mining. Las Vegas, USA: IEEE Press, 2007: 66-72.
［13］ Batista G E, Prati R C, Monard M C. A study of the behavior of several methods for balancing machine learning training data［J］. ACM SIGKDD Explorations Newsletter, 2004, 6(1): 20-29.
［14］ Weiss G M, Provost F J. Learning when training data are costly: The effect of class distribution on tree induction［J］. Journal of Artificial Intelligence Research, 2003, 19: 315-354.
［15］ Domingos P. MetaCost: A general method for making classifiers cost-sensitive［C］// Proceedings of the International Conference on Knowledge Discovery and Data Mining. San Diego, USA: ACM Press, 1999: 155-164.
［16］ Chen C, Liaw A, Breiman L. Using random forest to learn imbalanced data［R］. TR666, Statistics Department, University of California at Berkeley, 2004.
［17］ Chew H G, Bogner R E, Lim C C. Dual ν-support vector machine with error rate and training size biasing［C］// Proceedings of the 26th International Conference on Acoustics, Speech and Signal Processing. Salt Lake City, USA: IEEE Press, 2001, 2: 1 269-1 272.
［18］ Raskutti B, Kowalczyk A. Extreme re-balancing for SVMs: A case study［J］. ACM SIGKDD Explorations Newsletter, 2004, 6(1): 60-69.
［19］ Juszczak P, Duin R P W. Uncertainty sampling methods for one-class classifiers［C］// Proceedings of International Conference on Machine Learning. Washington, USA: IEEE Press, 2003: 81-88.
［20］ Zhou Z H, Liu X Y. Training cost-sensitive neural networks with methods addressing the class imbalance problem［J］. IEEE Transactions on Knowledge and Data Engineering, 2006, 18(1): 63-77.
［21］ Liu X Y, Wu J X, Zhou Z H. Exploratory undersampling for class-imbalance learning［J］. IEEE Transactions on Systems, Man, and Cybernetics, 2009, 39(2): 539-550.
［22］ Galar M, Fernandez A, Barrenechea E, et al. A review on ensembles for the class imbalance problem: Bagging-, boosting-, and hybrid-based approaches［J］. IEEE Transactions on Systems, Man, and Cybernetics, 2012, 42(4): 463-484.
［23］ Viola P, Jones M. Rapid object detection using a boosted cascade of simple features［C］// Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. London, IEEE Press, 2001, 1: I-511-518.
［24］ Liu X Y, Li Q Q, Zhou Z H. Learning imbalanced multi-class data with optimal dichotomy weights［C］// IEEE 13th International Conference on Data Mining. Omaha, USA: IEEE Press, 2013: 478-487.
［25］ Drummond C, Holte R C. C4.5, class imbalance, and cost sensitivity: Why under-sampling beats over-sampling［EB/OL］. http://www.site.uottawa.ca/~nat/Workshop2003/drummondc.pdf.

[1]	马玉莲, 崔文泉. 一种预测高频价格的端到端双目标多任务方法[J]. 中国科学技术大学学报, 2021, 51(3): 246-258.
[2]	高翔，陈力. 分组随机梯度下降法：掉队和延迟的平衡[J]. 中国科学技术大学学报, 2020, 50(8): 1156-1161.
[3]	李鹏，郑宇，张谈贵. 基于视觉显著性的无人机目标跟踪[J]. 中国科学技术大学学报, 2020, 50(8): 1162-1169.
[4]	李永军，曹为华，凌强. 基于特征点轨迹的多目标跟踪算法[J]. 中国科学技术大学学报, 2020, 50(6): 726-732.
[5]	李健聪，王泰然，舒武，胡素磊，欧阳润海，李微雪. 基于人工智能的描述符预测合金催化材料形成能[J]. 中国科学技术大学学报, 2020, 50(6): 844-851.
[6]	龙飞, 黄琨, 李峰. 一种机器学习与相变之间的新型映射[J]. 中国科学技术大学学报, 2020, 50(1): 18-28.
[7]	刘雁，饶元. 基于多领域复杂网络拓扑结构的节点重要度评价方法[J]. 中国科学技术大学学报, 2019, 49(7): 533-543.
[8]	宋辉，杨明. 基于辅助信息的混合线性矩阵补全模型[J]. 中国科学技术大学学报, 2019, 49(7): 572-578.
[9]	尹浩，刘桂建，钮志远，冯艾荣. 利用真空热解炉—质谱联用仪研究不同电子电离能量下淮南烟煤和无烟煤热解产物的分子组成分布特征[J]. 中国科学技术大学学报, 2019, 49(4): 311-320.
[10]	宋辉，杨明. 基于辅助信息的混合线性矩阵补全模型[J]. 中国科学技术大学学报, 2019, 49(2): 159-165.
[11]	郝焕焕. 非负张量和超图的主特征向量[J]. 中国科学技术大学学报, 2019, 49(11): 897-901.
[12]	汪中，陈恩红，刘贵全. 基于特征融合的二代身份证人脸验证系统[J]. 中国科学技术大学学报, 2019, 49(10): 828-834.
[13]	谭嘉莉，何钰，吴燕晶，孙广中. 应用于校园心理咨询的对话匹配度预测模型[J]. 中国科学技术大学学报, 2018, 48(9): 739-747.
[14]	李勇明，肖洁，王品，颜芳. 基于封装式偏差回归的心脏生理和病理年龄估计算法[J]. 中国科学技术大学学报, 2018, 48(9): 762-769.
[15]	严菲，王晓栋. 基于自适应局部保持投影的无监督特征选择方法[J]. 中国科学技术大学学报, 2018, 48(4): 290-297.

一种基于类别不平衡数据的层次分类模型

A hierarchical classification model for class-imbalanced data

PDF (PC)

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

本文评价