基于排序互信息的无监督特征选择

doi:10.3969/j.issn.1000-1565.2020.02.013

河北大学学报(自然科学版) ›› 2020, Vol. 40 ›› Issue (2): 200-204.DOI: 10.3969/j.issn.1000-1565.2020.02.013

基于排序互信息的无监督特征选择

李纯果^1,2,张春琴¹,李海峰³

收稿日期:2019-09-20 出版日期:2020-03-25 发布日期:2020-03-25
通讯作者: 李海峰(1980—),男,河北唐县人,河北大学副教授,主要从事数据挖掘、信息分析等方面研究.E-mail:hbulhf@126.com
作者简介:李纯果(1981—),女,河北邯郸人,河北大学副教授,博士,主要从事统计学习、无监督学习排序、信息论、无监督评价等方面的研究. E-mail: licg@hbu.cn
基金资助:
国家自然科学基金资助项目(61573348);河北省教育厅资助项目(ZC2016157)

Unsupervised feature selection based on ranking mutual information

LI Chunguo^1,2,ZHANG Chunqin¹,LI Haifeng³

1.College of Mathematics and Information Science, Hebei University, Baoding 071002, China; 2.Hebei Key Laboratory of Machine Learning and Computational Intelligence, Baoding 071002, China; 3.Department of Computer Teaching, Hebei University, Baoding 071002, China

Received:2019-09-20 Online:2020-03-25 Published:2020-03-25

摘要/Abstract

摘要： 根据排序问题的单调先验知识,无监督学习问题中的观测属性之间也具备单调关系;否则该属性与排序无关,为冗余属性.基于排序互信息反应的两属性之间的单调关系,提出用每个属性与其他属性之间的平均互信息,来衡量每个属性与排序学习的相关程度,具有最高的平均互信息即为排序最相关的属性.

关键词: 无监督排序, 特征选择, 排序互信息

Abstract: Based on ranking prior knowledge of monotonicity,each observation attribute should be monotonic with the other observation attributes for unsupervised ranking problems.Otherwise,the attribute would be irrelevant with ranking and should be assumed to a redundant attribute.Based on the ranking mutual information,which reflects the monotonic degree between observation attributes and the order sequence,mean ranking mutual information is proposed to measure the monotonicity between observation attributes.The most relevant attributes should be with the biggest ranking mutual information.

Key words: unsupervised ranking, feature selection, ranking mutual information

中图分类号:

O235

李纯果,张春琴,李海峰. 基于排序互信息的无监督特征选择[J]. 河北大学学报(自然科学版), 2020, 40(2): 200-204.

LI Chunguo,ZHANG Chunqin,LI Haifeng. Unsupervised feature selection based on ranking mutual information[J]. Journal of Hebei University (Natural Science Edition), 2020, 40(2): 200-204.

参考文献

[1] 李纯果,李海峰.无监督排序学习算法的一致性比较[J].河北大学学报(自然科学版),2015,35(2):182-187.DOI: 10.3969/j.issn.1000-1565.2015.02.013.
[2] YU B.Research on information retrieval model based on ontology[J].EURASIP Journal on Wireless Communications and Networking,2019,2019(30)1-8.DOI:10.1186/s13638-019-1354-z.
[3] MA S,GONG C,HU R,et al.Query independent scholarly article ranking[C] //2018 IEEE 34th International Conference on Data Engineering(ICDE),2018:953-964.DOI:10.1109/ICDE.2018.00090.
[4] LAI Y A,HSU C C,CHEN W H,et al.DeepRank: improving unsupervised node ranking via link discovery[J].Data Mining and Knowledge Discovery,2019,33: 474–498.DOI:10.1007/s10618-018-0601-y.
[5] 李纯果.基于主曲线的无监督排序学习及其在综合评价中的应用[D].北京:中国科学院自动化研究所,2015.
[6] 李海峰,李纯果.深度学习结构和算法比较分析[J].河北大学学报(自然科学版),2012,32(5):538-544.DOI: 10.3969/j.issn.1000-1565.2012.05.016.
[7] SCHOLKOPF B,SIMARD P,SMOLA AJ,et al.Prior knowledge in support vector kernels[C] //Twelfth Conference on Neural Information processing Systems,1998:640-646.
[8] VAPNIK V,VASHIST A.A new learning paradigm: learning using privileged information[J].Neural Networks,2009,22(5): 544-557.DOI: 10.1016/j.neunet.2009.06.042.
[9] 胡包钢,王泳,杨双红,等.如何增加人工神经元网络的透明度[J].模式识别与人工智能,2007,20(1): 72-84.DOI: 10.3969/j.issn.1003-6059.2007.01.012.
[10] HU Q H,GUO M Z,YU D R,et al.Information entropy for ordinal classification[J].Science China Information Sciences,2010,53(6): 1188-1200.DOI: 10.1007/s11432-010-3117-7.
[11] HU Q H,CHE X,ZHANG L.Rank entropy-based decision trees for monotonic classification[J].IEEE Transactions on Knowledge and Data Engineering,2012,24(11): 2052-2064.DOI: 10.1109/tkde.2011.149.
[12] 张文修,仇国芳.基于粗糙集的不确定决策[M].北京:清华大学出版社,2005.

基于排序互信息的无监督特征选择

Unsupervised feature selection based on ranking mutual information

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 2

编辑推荐

Metrics

本文评价

[1]	张群峰. 基于宽松下近似的模糊决策树归纳算法[J]. 河北大学学报(自然科学版), 2016, 36(3): 225-228.
[2]	张群峰,张天一. 基于宽松下近似概念的连续型决策表的属性约简方法[J]. 河北大学学报(自然科学版), 2013, 33(4): 337-341.