Feature selection algorithm for imbalanced data based on pseudo-label consistency

doi:10.11772/j.issn.1001-9081.2021050957

Journal of Computer Applications ›› 2022, Vol. 42 ›› Issue (2): 475-484.DOI: 10.11772/j.issn.1001-9081.2021050957

• Data science and technology • Previous Articles Next Articles

Feature selection algorithm for imbalanced data based on pseudo-label consistency

Yiheng LI, Chenxi DU, Yanyan YANG(), Xiangyu LI

School of Software Engineering，Beijing Jiaotong University，Beijing 100044，China

Received:2021-03-25 Revised:2021-07-21 Accepted:2021-07-21 Online:2022-02-11 Published:2022-02-10
Contact: Yanyan YANG
About author:LI Yiheng， born in 2001. His research interests include machine learning.
DU Chenxi， born in 2001. His research interests include machine learning.
YANG Yanyan， born in 1986， Ph. D.， lecturer. Her research interests include machine learning， granular computing， rough set.
LI Xiangyu， born in 1990， Ph. D.， lecturer. Her research interests include bioinformatics， machine learning.
Supported by:
National Natural Science Foundation of China(61806108);Fundamental Research Funds for the Central Universities(2019RC055);Beijing Training Program of Innovation for College Students(202110004107)

基于伪标签一致度的不平衡数据特征选择算法

李懿恒, 杜晨曦, 杨燕燕(), 李翔宇

北京交通大学软件学院，北京 100044

通讯作者: 杨燕燕
作者简介:李懿恒（2001—），男，山西临汾人，主要研究方向：机器学习；
杜晨曦（2001—），男，吉林梅河口人，主要研究方向：机器学习；
杨燕燕（1986—），女，河南郑州人，讲师，博士，CCF会员，主要研究方向：机器学习、粒计算、粗糙集；
李翔宇（1990—），女，河南郑州人，讲师，博士，主要研究方向：生物信息学、机器学习。
基金资助:
国家自然科学基金资助项目(61806108);中央高校基本科研业务费专项资金资助项目(2019RC055);北京市级大学生创新创业训练计划项目(202110004107)

Abstract

Abstract:

Aiming at the problem that most algorithms of granular computing ignore the class-imbalance of data， a feature selection algorithm integrating pseudo-label strategy was proposed to deal with class-imbalanced data. Firstly， to investigate feature selection from class-imbalanced data conveniently， the sample consistency and dataset consistency were re-defined， and the corresponding greedy forward search algorithm for feature selection was designed. Then， the pseudo-label strategy was introduced to balance the class distribution of the data. By integrating the learned pseudo-label of a sample into consistency measure， the pseudo-label consistency was defined to estimate the features of the class-imbalanced dataset. Finally， an algorithm for Pseudo-Label Consistency based Feature Selection （PLCFS） for class-imbalanced data was developed based on the preservation of the pseudo-label consistency measure for the class-imbalanced dataset. Experimental results indicate that the proposed PLCFS has the performance only lower than max-Relevancy and Min-Redundancy （mRMR） algorithm， but outperforms Relief algorithm and algorithm for Consistency-based Feature Selection （CFS）.

Key words: granular computing, pseudo-label, class-imbalanced data, feature selection, consistency measure

摘要：

针对大多数粒计算特征选择算法未考虑数据的类别不平衡性的问题，提出一种融合伪标签策略的类别不平衡数据特征选择算法。首先，为了便于研究类别不平衡数据特征选择算法，重新定义样本和数据集一致度的概念，并设计了相应特征选择的贪婪前向搜索算法；其次，引入伪标签策略以平衡数据的类别分布，并将所学样本的伪标签融入一致性测度中，以构造伪标签一致度来估计类别不平衡数据集的特征；最后，通过保持类别不平衡数据集的伪标签一致度不变，设计一种面向类别不平衡数据的基于伪标签一致性的特征选择算法（PLCFS）。实验结果表明，所提PLCFS的性能仅次于最大相关最小冗余（mRMR）算法，而优于Relief算法和基于一致性的特征选择算法（CFS）。

关键词: 粒计算, 伪标签, 类别不平衡数据, 特征选择, 一致性测度

CLC Number:

TP391

Yiheng LI, Chenxi DU, Yanyan YANG, Xiangyu LI. Feature selection algorithm for imbalanced data based on pseudo-label consistency[J]. Journal of Computer Applications, 2022, 42(2): 475-484.

李懿恒, 杜晨曦, 杨燕燕, 李翔宇. 基于伪标签一致度的不平衡数据特征选择算法[J]. 《计算机应用》唯一官方网站, 2022, 42(2): 475-484.

Figures/Tables 11

References 24

1	李艳霞，柴毅，胡友强，等.不平衡数据分类方法综述［J］.控制与决策， 2019， 34（4）： 673-688. 10.13195/j.kzyjc.2018.0865
	LI Y X， CHAI Y， HU Y Q， et al. Review of imbalanced data classification methods［J］. Control and Decision， 2019， 34（4）： 673-688. 10.13195/j.kzyjc.2018.0865
2	HE H B， GARCIA E A. Learning from imbalanced data［J］. IEEE Transactions on Knowledge and Data Engineering， 2009， 21（9）： 1263-1284. 10.1109/tkde.2008.239
3	JING X Y， ZHANG X Y， ZHU X K， et al. Multiset feature learning for highly imbalanced data classification［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2021， 43（1）： 139-156. 10.1109/tpami.2019.2929166
4	KHORSHIDI H A， AICKELIN U. Constructing classifiers for imbalanced data using diversity optimization［J］. Information Sciences， 2021， 565： 1-16. 10.1016/j.ins.2021.02.069
5	FU Y G， HUANG H Y， GUAN Y， et al. EBRB cascade classifier for imbalanced data via rule weight updating［J］. Knowledge-Based Systems， 2021， 223： No.107010. 10.1016/j.knosys.2021.107010
6	ZHENG Z H， WU X Y， SRIHARI R. Feature selection for text categorization on imbalanced data［J］. ACM SIGKDD Explorations Newsletter， 2004， 6（1）： 80-89. 10.1145/1007730.1007741
7	HULSE J VAN， KHOSHGOFTAAR T M， NAPOLITANO A， et al. Feature selection with high-dimensional imbalanced data ［C］// Proceedings of the 2009 IEEE International Conference on Data Mining Workshops. Piscataway： IEEE， 2009： 507-514. 10.1109/icdmw.2009.35
8	WASIKOWSKI M， CHEN X W. Combating the small sample class imbalance problem using feature selection［J］. IEEE Transactions on Knowledge and Data Engineering， 2010， 22（10）： 1388-1400. 10.1109/tkde.2009.187
9	MALDONADO S， WEBER R， FAMILI F. Feature selection for high-dimensional class-imbalanced data sets using Support Vector Machines［J］. Information Sciences， 2014， 286： 228-246. 10.1016/j.ins.2014.07.015
10	YIN L Z， GE Y， XIAO K L， et al. Feature selection for high-dimensional imbalanced data［J］. Neurocomputing， 2013， 105： 3-11. 10.1016/j.neucom.2012.04.039
11	FU G H， WU Y J， ZONG M J， et al. Feature selection and classification by minimizing overlap degree for class-imbalanced data in metabolomics［J］. Chemometrics and Intelligent Laboratory Systems， 2020， 196： No.103906. 10.1016/j.chemolab.2019.103906
12	PEDRYCZ W. Granular Computing： Analysis and Design of Intelligent Systems［M］. Boca Raton： CRC Press， 2013： 15-36. 10.1201/b14862
13	YAO Y Y. Three-way granular computing， rough sets， and formal concept analysis［J］. International Journal of Approximate Reasoning， 2020， 116： 106-125. 10.1016/j.ijar.2019.11.002
14	LIU J F， HU Q H， YU D R. A weighted rough set based method developed for class imbalance learning［J］. Information Sciences， 2008， 178（4）： 1235-1256. 10.1016/j.ins.2007.10.002
15	ZHOU P， HU X G， LI P P， et al. Online feature selection for high-dimensional class-imbalanced data［J］. Knowledge-Based Systems， 2017， 136： 187-199. 10.1016/j.knosys.2017.09.006
16	CHEN H M， LI T R， FAN X， et al. Feature selection for imbalanced data based on neighborhood rough sets［J］. Information Sciences， 2019， 483： 1-20. 10.1016/j.ins.2019.01.041
17	XUE J H， HALL P. Why does rebalancing class-unbalanced data improve AUC for linear discriminant analysis？［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2015， 37（5）： 1109-1112. 10.1109/tpami.2014.2359660
18	YANG Y Z， XU Z. Rethinking the value of labels for improving class-imbalanced learning［C/OL］// Proceedings of the 34th Conference on Neural Information Processing Systems. ［2021-03-28］. .
19	YANG X B， LIANG S C， YU H L， et al. Pseudo-label neighborhood rough set： measures and attribute reductions［J］. International Journal of Approximate Reasoning， 2019， 105： 112-129. 10.1016/j.ijar.2018.11.010
20	ZENG W R， CHEN X W， CHENG H. Pseudo labels for imbalanced multi-label learning ［C］// Proceedings of the 2014 International Conference on Data Science and Advanced Analytics. Piscataway： IEEE， 2014： 25-31. 10.1109/dsaa.2014.7058047
21	MIAO D Q， ZHAO Y， YAO Y Y， et al. Relative reducts in consistent and inconsistent decision tables of the Pawlak rough set model［J］. Information Sciences， 2009， 179（24）： 4140-4150. 10.1016/j.ins.2009.08.020
22	YANG Y Y， CHEN D G， WANG H. Active sample selection based incremental algorithm for attribute reduction with rough sets［J］. IEEE Transactions on Fuzzy Systems， 2017， 25（4）： 825-838. 10.1109/tfuzz.2016.2581186
23	PENG H C， LONG F H， DING C. Feature selection based on mutual information criteria of max-dependency， max-relevance， and min-redundancy［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2005， 27（8）： 1226-1238. 10.1109/tpami.2005.159
24	KONONENKO I. Estimating attributes： analysis and extensions of RELIEF ［C］// Proceedings of the 1994 European Conference on Machine Learning， LNCS784. Berlin： Springer， 1994： 171-182.

序号	数据集	I	F	IR	P/%	N/%	数据类型
D1	arrhythmia	452	279	9.27	9.73	90.27	mixed
D2	crx	690	15	1.25	44.49	55.51	mixed
D3	glass	214	9	2.06	32.71	67.29	numerical
D4	heart	270	13	1.25	44.44	55.56	mixed
D5	segmentation	2 308	19	6.02	14.25	85.75	numerical
D6	tic-tac-toe	958	9	1.89	34.66	65.34	nominal
D7	wdbc	569	30	1.68	37.26	62.74	numerical
D8	wpbc	198	33	3.21	23.74	76.26	numerical
D9	yeast	1 484	8	2.46	28.91	71.09	numerical
D10	zoo	101	16	19.20	4.95	95.05	nominal

序号	数据集	I	F	IR	P/%	N/%	数据类型
D1	arrhythmia	452	279	9.27	9.73	90.27	mixed
D2	crx	690	15	1.25	44.49	55.51	mixed
D3	glass	214	9	2.06	32.71	67.29	numerical
D4	heart	270	13	1.25	44.44	55.56	mixed
D5	segmentation	2 308	19	6.02	14.25	85.75	numerical
D6	tic-tac-toe	958	9	1.89	34.66	65.34	nominal
D7	wdbc	569	30	1.68	37.26	62.74	numerical
D8	wpbc	198	33	3.21	23.74	76.26	numerical
D9	yeast	1 484	8	2.46	28.91	71.09	numerical
D10	zoo	101	16	19.20	4.95	95.05	nominal

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.555 0	0.541 8	0.6592	0.550 6	0.100 1	0.073 1	0.2378	0.102 4	0.725 3	0.715 8	0.7968	0.722 0
D2	0.823 2	0.824 6	0.815 9	0.8464	0.686 1	0.687 6	0.678 6	0.7132	0.823 2	0.824 6	0.815 9	0.8464
D3	0.614 3	0.614 3	0.6422	0.614 3	0.489 1	0.490 9	0.5289	0.488 6	0.492 6	0.492 3	0.5114	0.492 6
D4	0.8481	0.840 7	0.837 0	0.833 3	0.8448	0.837 1	0.833 5	0.830 5	0.8481	0.840 7	0.837 0	0.833 3
D5	0.781 0	0.776 2	0.7831	0.773 2	0.6979	0.665 9	0.695 7	0.683 9	0.864 7	0.862 1	0.8663	0.859 5
D6	—	0.6864	0.686 4	0.660 3	—	0.4346	0.434 6	0.429 6	—	0.6864	0.686 4	0.660 3
D7	0.949 0	0.9490	0.9490	0.933 2	0.942 5	0.941 9	0.9430	0.925 2	0.949 0	0.949 0	0.9490	0.933 2
D8	0.759 0	0.7590	0.7590	0.759 0	0.688 6	0.688 6	0.6886	0.688 6	0.159 0	0.159 0	0.1590	0.159 0
D9	—	0.476 4	0.4778	0.477 8	—	0.227 9	0.2679	0.228 8	—	0.669 5	0.6695	0.669 5
D10	0.9000	0.840 5	0.890 0	0.900 0	0.8174	0.680 7	0.803 5	0.809 3	0.736 7	0.8865	0.731 7	0.738 5

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.555 0	0.541 8	0.6592	0.550 6	0.100 1	0.073 1	0.2378	0.102 4	0.725 3	0.715 8	0.7968	0.722 0
D2	0.823 2	0.824 6	0.815 9	0.8464	0.686 1	0.687 6	0.678 6	0.7132	0.823 2	0.824 6	0.815 9	0.8464
D3	0.614 3	0.614 3	0.6422	0.614 3	0.489 1	0.490 9	0.5289	0.488 6	0.492 6	0.492 3	0.5114	0.492 6
D4	0.8481	0.840 7	0.837 0	0.833 3	0.8448	0.837 1	0.833 5	0.830 5	0.8481	0.840 7	0.837 0	0.833 3
D5	0.781 0	0.776 2	0.7831	0.773 2	0.6979	0.665 9	0.695 7	0.683 9	0.864 7	0.862 1	0.8663	0.859 5
D6	—	0.6864	0.686 4	0.660 3	—	0.4346	0.434 6	0.429 6	—	0.6864	0.686 4	0.660 3
D7	0.949 0	0.9490	0.9490	0.933 2	0.942 5	0.941 9	0.9430	0.925 2	0.949 0	0.949 0	0.9490	0.933 2
D8	0.759 0	0.7590	0.7590	0.759 0	0.688 6	0.688 6	0.6886	0.688 6	0.159 0	0.159 0	0.1590	0.159 0
D9	—	0.476 4	0.4778	0.477 8	—	0.227 9	0.2679	0.228 8	—	0.669 5	0.6695	0.669 5
D10	0.9000	0.840 5	0.890 0	0.900 0	0.8174	0.680 7	0.803 5	0.809 3	0.736 7	0.8865	0.731 7	0.738 5

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.537 3	0.526 4	0.6659	0.546 2	0.120 5	0.130 6	0.2786	0.182 8	0.714 4	0.707 4	0.8020	0.719 9
D2	0.829 0	0.826 1	0.8536	0.826 1	0.707 2	0.704 4	0.7405	0.694 0	0.829 0	0.826 1	0.8536	0.826 1
D3	0.646 8	0.646 8	0.651 5	0.6561	0.475 7	0.494 7	0.468 0	0.4988	0.526 3	0.525 8	0.530 9	0.5322
D4	0.8333	0.811 1	0.814 8	0.800 0	0.8299	0.805 7	0.811 9	0.795 4	0.8333	0.811 1	0.814 8	0.800 0
D5	0.822 5	0.826 0	0.8290	0.819 0	0.701 7	0.7043	0.703 8	0.695 6	0.891 6	0.893 7	0.8964	0.889 5
D6	—	0.690 6	0.717 8	0.7513	—	0.445 3	0.459 4	0.4909	—	0.690 6	0.7731	0.717 8
D7	0.947 3	0.940 2	0.9526	0.947 3	0.940 6	0.937 1	0.9476	0.940 7	0.947 3	0.943 8	0.9526	0.947 3
D8	0.663 3	0.6887	0.678 6	0.658 6	0.361 1	0.3855	0.381 2	0.375 5	0.663 3	0.6887	0.678 6	0.658 6
D9	—	0.425 8	0.4556	0.423 2	—	0.213 3	0.2173	0.213 1	—	0.630 1	0.6528	0.628 1
D10	0.860 0	0.820 5	0.8900	0.880 0	0.686 0	0.666 9	0.802 0	0.8065	0.710 3	0.8754	0.730 1	0.722 4

Feature selection algorithm for imbalanced data based on pseudo-label consistency

基于伪标签一致度的不平衡数据特征选择算法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 11

References 24

Related Articles 15

Recommended Articles

Metrics

数据集	CFS	PLCFS	mRMR	Relief
D1	19	18	24	24
D2	13	13	10	12
D3	8	7	5	7
D4	12	11	10	10
D5	13	16	13	14
D6	10	8	8	7
D7	14	24	24	12
D8	15	16	17	11
D9	9	7	8	7
D10	10	11	14	13

数据集	CFS	PLCFS	mRMR	Relief
D1	19	18	24	24
D2	13	13	10	12
D3	8	7	5	7
D4	12	11	10	10
D5	13	16	13	14
D6	10	8	8	7
D7	14	24	24	12
D8	15	16	17	11
D9	9	7	8	7
D10	10	11	14	13

[1]	Hong CHEN, Bing QI, Haibo JIN, Cong WU, Li’ang ZHANG. Class-imbalanced traffic abnormal detection based on 1D-CNN and BiGRU [J]. Journal of Computer Applications, 2024, 44(8): 2493-2499.
[2]	Xun YAO, Zhongzheng QIN, Jie YANG. Generative label adversarial text classification model [J]. Journal of Computer Applications, 2024, 44(6): 1781-1785.
[3]	Mingzhu LEI, Hao WANG, Rong JIA, Lin BAI, Xiaoying PAN. Oversampling algorithm based on synthesizing minority class samples using relationship between features [J]. Journal of Computer Applications, 2024, 44(5): 1428-1436.
[4]	Lin GAO, Yu ZHOU, Tak Wu KWONG. Evolutionary bi-level adaptive local feature selection [J]. Journal of Computer Applications, 2024, 44(5): 1408-1414.
[5]	Lin SUN, Menghan LIU. K-means clustering based on adaptive cuckoo optimization feature selection [J]. Journal of Computer Applications, 2024, 44(3): 831-841.
[6]	Dapeng XU, Xinmin HOU. Feature selection method for graph neural network based on network architecture design [J]. Journal of Computer Applications, 2024, 44(3): 663-670.
[7]	Shengjie MENG, Wanjun YU, Ying CHEN. Feature selection algorithm for high-dimensional data with maximum correlation and maximum difference [J]. Journal of Computer Applications, 2024, 44(3): 767-771.
[8]	Jingxin LIU, Wenjing HUANG, Liangsheng XU, Chong HUANG, Jiansheng WU. Unsupervised feature selection model with dictionary learning and sample correlation preservation [J]. Journal of Computer Applications, 2024, 44(12): 3766-3775.
[9]	Zhaoze GAO, Xiaofei ZHU, Nengqiang XIANG. Semi-supervised stance detection based on category-aware curriculum learning [J]. Journal of Computer Applications, 2024, 44(10): 3281-3287.
[10]	Tian HE, Zongxin SHEN, Qianqian HUANG, Yanyong HUANG. Adaptive learning-based multi-view unsupervised feature selection method [J]. Journal of Computer Applications, 2023, 43(9): 2657-2664.
[11]	Lin SUN, Jinxu HUANG, Jiucheng XU. Feature selection for imbalanced data based on neighborhood tolerance mutual information and whale optimization algorithm [J]. Journal of Computer Applications, 2023, 43(6): 1842-1854.
[12]	Yuanjiang LI, Jinsheng QUAN, Yangyi TAN, Tian YANG. Attribute reduction for high-dimensional data based on bi-view of similarity and difference [J]. Journal of Computer Applications, 2023, 43(5): 1467-1472.
[13]	Zhenhua YU, Zhengqi LIU, Ying LIU, Cheng GUO. Feature selection method based on self-adaptive hybrid particle swarm optimization for software defect prediction [J]. Journal of Computer Applications, 2023, 43(4): 1206-1213.
[14]	Lin SUN, Tianjiao MA, Zhan’ao XUE. Multilabel feature selection algorithm based on Fisher score and fuzzy neighborhood entropy [J]. Journal of Computer Applications, 2023, 43(12): 3779-3789.
[15]	Jingcheng XU, Xuebin CHEN, Yanling DONG, Jia YANG. DDoS attack detection by random forest fused with feature selection [J]. Journal of Computer Applications, 2023, 43(11): 3497-3503.

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.566 1	0.539 7	0.661 4	0.550 6	0.210 5	0.184 8	0.336 3	0.222 4	0.734 6	0.716 3	0.798 5	0.720 7
D2	0.810 1	0.821 7	0.834 8	0.826 4	0.691 9	0.706 1	0.717 9	0.713 2	0.810 1	0.821 7	0.834 8	0.797 1
D3	0.646 8	0.698 7	0.651 5	0.654 3	0.499 1	0.536 9	0.616 0	0.568 6	0.523 4	0.556 2	0.323 0	0.360 9
D4	0.814 8	0.818 5	0.814 8	0.813 3	0.807 9	0.812 4	0.809 4	0.807 5	0.814 8	0.818 5	0.814 8	0.814 8
D5	0.829 9	0.833 3	0.843 7	0.833 2	0.723 4	0.724 9	0.734 8	0.723 9	0.896 4	0.898 6	0.904 9	0.894 7
D6	—	0.593 2	0.498 4	0.560 3	—	0.362 6	0.322 2	0.329 6	—	0.593 2	0.498 4	0.745 0
D7	0.935 0	0.943 8	0.938 5	0.935 0	0.926 2	0.937 3	0.930 5	0.924 2	0.935 0	0.943 8	0.938 5	0.938 5
D8	0.633 2	0.648 5	0.658 8	0.641 1	0.363 9	0.357 4	0.378 7	0.369 3	0.633 2	0.648 5	0.658 8	0.593 6
D9	—	0.444 1	0.444 1	0.444 1	—	0.279 4	0.315 0	0.287 3	—	0.643 8	0.643 8	0.641 7
D10	0.940 0	0.880 5	0.920 0	0.900 0	0.857 4	0.650 0	0.789 8	0.755 5	0.763 4	0.919 7	0.749 5	0.762 5

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.566 1	0.539 7	0.661 4	0.550 6	0.210 5	0.184 8	0.336 3	0.222 4	0.734 6	0.716 3	0.798 5	0.720 7
D2	0.810 1	0.821 7	0.834 8	0.826 4	0.691 9	0.706 1	0.717 9	0.713 2	0.810 1	0.821 7	0.834 8	0.797 1
D3	0.646 8	0.698 7	0.651 5	0.654 3	0.499 1	0.536 9	0.616 0	0.568 6	0.523 4	0.556 2	0.323 0	0.360 9
D4	0.814 8	0.818 5	0.814 8	0.813 3	0.807 9	0.812 4	0.809 4	0.807 5	0.814 8	0.818 5	0.814 8	0.814 8
D5	0.829 9	0.833 3	0.843 7	0.833 2	0.723 4	0.724 9	0.734 8	0.723 9	0.896 4	0.898 6	0.904 9	0.894 7
D6	—	0.593 2	0.498 4	0.560 3	—	0.362 6	0.322 2	0.329 6	—	0.593 2	0.498 4	0.745 0
D7	0.935 0	0.943 8	0.938 5	0.935 0	0.926 2	0.937 3	0.930 5	0.924 2	0.935 0	0.943 8	0.938 5	0.938 5
D8	0.633 2	0.648 5	0.658 8	0.641 1	0.363 9	0.357 4	0.378 7	0.369 3	0.633 2	0.648 5	0.658 8	0.593 6
D9	—	0.444 1	0.444 1	0.444 1	—	0.279 4	0.315 0	0.287 3	—	0.643 8	0.643 8	0.641 7
D10	0.940 0	0.880 5	0.920 0	0.900 0	0.857 4	0.650 0	0.789 8	0.755 5	0.763 4	0.919 7	0.749 5	0.762 5

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.416 0	0.433 5	0.661 5	0.444 5	0.174 1	0.205 7	0.314 1	0.203 1	0.624 7	0.640 3	0.798 8	0.648 6
D2	0.791 3	0.811 6	0.821 7	0.800 0	0.691 9	0.718 7	0.696 2	0.687 5	0.791 3	0.811 6	0.821 7	0.800 0
D3	0.660 9	0.660 9	0.656 3	0.660 9	0.635 6	0.635 6	0.631 5	0.635 6	0.333 5	0.333 5	0.329 0	0.333 5
D4	0.751 9	0.785 2	0.807 4	0.781 5	0.747 0	0.781 6	0.804 4	0.778 2	0.751 9	0.785 2	0.807 4	0.781 5
D5	0.830 7	0.829 9	0.833 3	0.825 5	0.721 0	0.702 2	0.724 2	0.716 0	0.897 2	0.897 4	0.898 7	0.893 7
D6	—	0.565 5	0.557 1	0.758 6	—	0.377 4	0.375 2	0.491 3	—	0.565 5	0.557 1	0.758 6
D7	0.922 6	0.919 2	0.913 9	0.936 7	0.913 2	0.910 6	0.903 9	0.928 5	0.922 6	0.919 2	0.913 9	0.936 7
D8	0.533 2	0.542 9	0.573 1	0.517 8	0.339 0	0.328 5	0.336 2	0.315 6	0.533 2	0.542 9	0.573 1	0.517 8
D9	—	0.450 2	0.452 2	0.447 5	—	0.253 8	0.310 0	0.213 9	—	0.648 7	0.650 2	0.646 6
D10	0.960 0	0.890 5	0.930 0	0.960 0	0.924 4	0.700 0	0.862 9	0.894 5	0.774 7	0.923 7	0.753 5	0.775 5

数据集	micro-F1				macro-F1				G-Mean
数据集	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief	CFS	PLCFS	mRMR	Relief
D1	0.416 0	0.433 5	0.661 5	0.444 5	0.174 1	0.205 7	0.314 1	0.203 1	0.624 7	0.640 3	0.798 8	0.648 6
D2	0.791 3	0.811 6	0.821 7	0.800 0	0.691 9	0.718 7	0.696 2	0.687 5	0.791 3	0.811 6	0.821 7	0.800 0
D3	0.660 9	0.660 9	0.656 3	0.660 9	0.635 6	0.635 6	0.631 5	0.635 6	0.333 5	0.333 5	0.329 0	0.333 5
D4	0.751 9	0.785 2	0.807 4	0.781 5	0.747 0	0.781 6	0.804 4	0.778 2	0.751 9	0.785 2	0.807 4	0.781 5
D5	0.830 7	0.829 9	0.833 3	0.825 5	0.721 0	0.702 2	0.724 2	0.716 0	0.897 2	0.897 4	0.898 7	0.893 7
D6	—	0.565 5	0.557 1	0.758 6	—	0.377 4	0.375 2	0.491 3	—	0.565 5	0.557 1	0.758 6
D7	0.922 6	0.919 2	0.913 9	0.936 7	0.913 2	0.910 6	0.903 9	0.928 5	0.922 6	0.919 2	0.913 9	0.936 7
D8	0.533 2	0.542 9	0.573 1	0.517 8	0.339 0	0.328 5	0.336 2	0.315 6	0.533 2	0.542 9	0.573 1	0.517 8
D9	—	0.450 2	0.452 2	0.447 5	—	0.253 8	0.310 0	0.213 9	—	0.648 7	0.650 2	0.646 6
D10	0.960 0	0.890 5	0.930 0	0.960 0	0.924 4	0.700 0	0.862 9	0.894 5	0.774 7	0.923 7	0.753 5	0.775 5

数据集	CFS	PLCFS	mRMR	Relief
D1	19	18	24	24
D2	13	13	10	12
D3	8	7	5	7
D4	12	11	10	10
D5	13	16	13	14
D6	10	8	8	7
D7	14	24	24	12
D8	15	16	17	11
D9	9	7	8	7
D10	10	11	14	13

数据集	CFS	PLCFS	mRMR	Relief
D1	19	18	24	24
D2	13	13	10	12
D3	8	7	5	7
D4	12	11	10	10
D5	13	16	13	14
D6	10	8	8	7
D7	14	24	24	12
D8	15	16	17	11
D9	9	7	8	7
D10	10	11	14	13

数据集	CFS	PLCFS	mRMR	Relief
D1	19	18	24	24
D2	13	13	10	12
D3	8	7	5	7
D4	12	11	10	10
D5	13	16	13	14
D6	10	8	8	7
D7	14	24	24	12
D8	15	16	17	11
D9	9	7	8	7
D10	10	11	14	13

数据集	CFS	PLCFS	mRMR	Relief
D1	19	18	24	24
D2	13	13	10	12
D3	8	7	5	7
D4	12	11	10	10
D5	13	16	13	14
D6	10	8	8	7
D7	14	24	24	12
D8	15	16	17	11
D9	9	7	8	7
D10	10	11	14	13