基于邻域粗糙集和帝王蝶优化的特征选择算法

doi:10.11772/j.issn.1001-9081.2021030497

《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (5): 1355-1366.DOI: 10.11772/j.issn.1001-9081.2021030497

基于邻域粗糙集和帝王蝶优化的特征选择算法

孙林¹^,²(), 赵婧¹, 徐久成¹^,², 王欣雅¹

^1.河南师范大学计算机与信息工程学院, 河南新乡 453007
^2.教育人工智能与个性化学习河南省重点实验室(河南师范大学), 河南新乡 453007

收稿日期:2021-04-02 修回日期:2021-09-15 接受日期:2021-09-22 发布日期:2022-06-11 出版日期:2022-05-10
通讯作者: 孙林
作者简介:孙林（1979—），男，河南南阳人，副教授，博士，CCF会员，主要研究方向：粒计算、数据挖掘、机器学习、生物信息学 sunlin@htu.edu.cn
赵婧（1996—），女，河南洛阳人，硕士研究生，主要研究方向：数据挖掘、机器学习
徐久成（1963—），男，河南洛阳人，教授，博士生导师，博士，CCF高级会员，主要研究方向：粒计算、数据挖掘、机器学习
王欣雅（1997—），女，河南平顶山人，硕士研究生，主要研究方向：数据挖掘、机器学习。
基金资助:
国家自然科学基金资助项目(62076089);河南省科技攻关项目(212102210136)

Feature selection algorithm based on neighborhood rough set and monarch butterfly optimization

Lin SUN¹^,²(), Jing ZHAO¹, Jiucheng XU¹^,², Xinya WANG¹

^1.College of Computer and Information Engineering，Henan Normal University，Xinxiang Henan 453007，China
^2.Key Laboratory of Artificial Intelligence and Personalized Learning in Education of Henan Province （Henan Normal University），Xinxiang Henan 453007，China

Received:2021-04-02 Revised:2021-09-15 Accepted:2021-09-22 Online:2022-06-11 Published:2022-05-10
Contact: Lin SUN
About author:SUN Lin， born in 1979，Ph. D.，associate professor. His researchinterests include granular computing，data mining，machine learning，bioinformatics.
ZHAO Jing，born in 1996，M. S. candidate. Her research interestsinclude data mining，machine learning.
XU Jiucheng， born in 1963，Ph. D.，professor. His researchinterests include granular computing，data mining，machine learning.
WANG Xinya， born in 1997，M. S. candidate. Her researchinterests include data mining，machine learning.
Supported by:
National Natural Science Foundation of China(62076089);Key Scientific and Technological Project of Henan Province(212102210136)

摘要/Abstract

摘要：

针对经典的帝王蝶优化（MBO）算法不能很好地处理连续型数据，以及粗糙集模型对于大规模、高维复杂的数据处理能力不足等问题，提出了基于邻域粗糙集（NRS）和MBO的特征选择算法。首先，将局部扰动和群体划分策略与MBO算法结合，并构建传输机制以形成一种二进制MBO（BMBO）算法；其次，引入突变算子增强算法的探索能力，设计了基于突变算子的BMBO（BMBOM）算法；然后，基于NRS的邻域度构造适应度函数，并对初始化的特征子集的适应度值进行评估并排序；最后，使用BMBOM算法通过不断迭代搜索出最优特征子集，并设计了一种元启发式特征选择算法。在基准函数上评估BMBOM算法的优化性能，并在UCI数据集上评价所提出的特征选择算法的分类能力。实验结果表明，在5个基准函数上，BMBOM算法的最优值、最差值、平均值以及标准差明显优于MBO和粒子群优化（PSO）算法；在UCI数据集上，与基于粗糙集的优化特征选择算法、结合粗糙集与优化算法的特征选择算法、结合NRS与优化算法的特征选择算法、基于二进制灰狼优化的特征选择算法相比，所提特征选择算法在分类精度、所选特征数和适应度值这3个指标上表现良好，能够选择特征数少且分类精度高的最优特征子集。

关键词: 帝王蝶优化, 特征选择, 邻域粗糙集, 邻域依赖度, 二进制

Abstract:

The classical Monarch Butterfly Optimization （MBO） algorithm cannot handle continuous data well， and the rough set model cannot sufficiently process large-scale， high-dimensional and complex data. To address these problems， a new feature selection algorithm based on Neighborhood Rough Set （NRS） and MBO was proposed. Firstly， local disturbance， group division strategy and MBO algorithm were combined， and a transmission mechanism was constructed to form a Binary MBO （BMBO） algorithm. Secondly， the mutation operator was introduced to enhance the exploration ability of this algorithm， and a BMBO based on Mutation operator （BMBOM） algorithm was proposed. Then， a fitness function was developed based on the neighborhood dependence degree in NRS， and the fitness values of the initialized feature subsets were evaluated and sorted. Finally， the BMBOM algorithm was used to search the optimal feature subset through continuous iterations， and a meta-heuristic feature selection algorithm was designed. The optimization performance of the BMBOM algorithm was evaluated on benchmark functions， and the classification performance of the proposed feature selection algorithm was evaluated on UCI datasets. Experimental results show that， the proposed BMBOM algorithm is significantly better than MBO and Particle Swarm Optimization （PSO） algorithms in terms of the optimal value， worst value， average value and standard deviation on five benchmark functions. Compared with the optimized feature selection algorithms based on rough set， the feature selection algorithms combining rough set and optimization algorithms， the feature selection algorithms combining NRS and optimization algorithms， the feature selection algorithms based on binary grey wolf optimization， the proposed feature selection algorithm performs well in the three indicators of classification accuracy， the number of selected features and fitness value on UCI datasets， and can select the optimal feature subset with few features and high classification accuracy.

Key words: Monarch Butterfly Optimization (MBO), feature selection, Neighborhood Rough Set (NRS), neighborhood dependence degree, binary

中图分类号:

TP181

孙林, 赵婧, 徐久成, 王欣雅. 基于邻域粗糙集和帝王蝶优化的特征选择算法[J]. 计算机应用, 2022, 42(5): 1355-1366.

Lin SUN, Jing ZHAO, Jiucheng XU, Xinya WANG. Feature selection algorithm based on neighborhood rough set and monarch butterfly optimization[J]. Journal of Computer Applications, 2022, 42(5): 1355-1366.

图/表 20

表1 六个基准函数信息

Tab. 1 Information of six benchmark functions

函数	公式	变量范围
Sphere	$f 1 (x) = ∑ i = 1 d x i 2$	［-5.12，5.12］ ^D
Schwefel2. 22	$f 2 (x) = ∑ i = 1 d x i + ∏ i = 1 d x i$	［-10，10］ ^D
Step	$f 3 (x) = ∑ i = 1 d x i + 0.5 2$	［-100，100］ ^D
Griank	$f 4 (x) = 1 4 000 ∑ i = 1 d x i 2 - ∏ i = 1 d c o s x i / i + 1$	［-600，600］ ^D
Ackley	$f 5 (x) = - 20 e x p - 15 1 d ∑ i = 1 d x i 2 - e x p 1 d ∑ i = 1 d c o s (2 π x i) + 20 + e$	［-30，30］ ^D
Rastrigin	$f 6 (x) = ∑ i = 1 d (x i 2 - 10 c o s (2 π x i) + 10)$	［-5.12，5.12］ ^D

表1 六个基准函数信息

Tab. 1 Information of six benchmark functions

函数	公式	变量范围
Sphere	$f 1 (x) = ∑ i = 1 d x i 2$	［-5.12，5.12］ ^D
Schwefel2. 22	$f 2 (x) = ∑ i = 1 d x i + ∏ i = 1 d x i$	［-10，10］ ^D
Step	$f 3 (x) = ∑ i = 1 d x i + 0.5 2$	［-100，100］ ^D
Griank	$f 4 (x) = 1 4 000 ∑ i = 1 d x i 2 - ∏ i = 1 d c o s x i / i + 1$	［-600，600］ ^D
Ackley	$f 5 (x) = - 20 e x p - 15 1 d ∑ i = 1 d x i 2 - e x p 1 d ∑ i = 1 d c o s (2 π x i) + 20 + e$	［-30，30］ ^D
Rastrigin	$f 6 (x) = ∑ i = 1 d (x i 2 - 10 c o s (2 π x i) + 10)$	［-5.12，5.12］ ^D

表2 三种优化算法在6个基准函数30维上的实验结果

Tab. 2 Experimental results of three optimization algorithms on 30-dimensions of six benchmark functions

函数	算法	最优值	最差值	平均值	标准差	运行时间/s
f₁	PSO	4.23E $+ 01$	6.12E+01	5.09E+01	4.06E+00	18
	MBO	7.90E $- 06$	2.15E+02	6.67E+01	7.92E+01	177
	BMBOM	4.75E-08	8.19E-05	8.66E-06	1.62E-05	516
f₂	PSO	7.76E+03	1.57E+04	1.20E+04	1.99E+03	54
	MBO	3.91E+02	5.06E+04	2.37E+04	1.49E+04	142
	BMBOM	0	1.36E+02	5.72E+01	4.58E+01	430
f₃	PSO	9.33E+03	1.60E+04	1.24E+04	1.63E+03	53
	MBO	0	7.28E+04	2.38E+04	2.71E+04	176
	BMBOM	0	8.28E+02	2.76E+01	1.51E+02	388
f₄	PSO	6.57E+05	1.43E+06	1.15E+06	1.66E+05	28
	MBO	1	6.36E+02	1.80E+02	2.29E+02	175
	BMBOM	1	3.18E+01	2.03E+00	5.62E+00	395
f₅	PSO	4.22E+01	1.35E+02	6.09E+01	2.72E+01	19
	MBO	1.32E-01	8.95E+01	3.46E+01	2.36E+01	138
	BMBOM	1.30E-04	9.65E-03	1.92E-03	2.16E-03	341
f₆	PSO	2.06E+02	2.66E+02	2.50E+02	1.37E+01	54
	MBO	2.32E-05	4.45E+02	1.68E+02	1.45E+02	177
	BMBOM	2.81E-07	1.12E+02	6.34E+00	2.27E+01	363

表2 三种优化算法在6个基准函数30维上的实验结果

Tab. 2 Experimental results of three optimization algorithms on 30-dimensions of six benchmark functions

函数	算法	最优值	最差值	平均值	标准差	运行时间/s
f₁	PSO	4.23E $+ 01$	6.12E+01	5.09E+01	4.06E+00	18
	MBO	7.90E $- 06$	2.15E+02	6.67E+01	7.92E+01	177
	BMBOM	4.75E-08	8.19E-05	8.66E-06	1.62E-05	516
f₂	PSO	7.76E+03	1.57E+04	1.20E+04	1.99E+03	54
	MBO	3.91E+02	5.06E+04	2.37E+04	1.49E+04	142
	BMBOM	0	1.36E+02	5.72E+01	4.58E+01	430
f₃	PSO	9.33E+03	1.60E+04	1.24E+04	1.63E+03	53
	MBO	0	7.28E+04	2.38E+04	2.71E+04	176
	BMBOM	0	8.28E+02	2.76E+01	1.51E+02	388
f₄	PSO	6.57E+05	1.43E+06	1.15E+06	1.66E+05	28
	MBO	1	6.36E+02	1.80E+02	2.29E+02	175
	BMBOM	1	3.18E+01	2.03E+00	5.62E+00	395
f₅	PSO	4.22E+01	1.35E+02	6.09E+01	2.72E+01	19
	MBO	1.32E-01	8.95E+01	3.46E+01	2.36E+01	138
	BMBOM	1.30E-04	9.65E-03	1.92E-03	2.16E-03	341
f₆	PSO	2.06E+02	2.66E+02	2.50E+02	1.37E+01	54
	MBO	2.32E-05	4.45E+02	1.68E+02	1.45E+02	177
	BMBOM	2.81E-07	1.12E+02	6.34E+00	2.27E+01	363

表3 14个UCI数据集信息

Tab. 3 Information of fourteen UCI datasets

序号	名称	特征数	样本数
1	Breastcancer	9	699
2	Congress	16	435
3	Heart	13	270
4	Hepatitis	19	155
5	Ionosphere	34	351
6	Lungcancer	56	32
7	Lymphography	18	148
8	Sonar	60	208
9	Spect	22	267
10	Tic-tac-toe	9	958
11	Vote	16	300
12	Waveform	40	5 000
13	Wine	13	178
14	Zoo	16	101

表4 FSNB与5种基于粗糙集的优化特征选择算法的分类精度实验结果 ( %)

Tab. 4 Experimental results of FSNB and five optimized feature selection algorithms based on rough sets on classification accuracy

数据集	FSNB	ARRSFA	PSORSFS	FSARSR	QCSIA_FS	DQBGOA_MR
Average	94.60	81.34	81.07	80.62	80.82	83.04
Breastcancer	95.67	92.61	92.30	91.28	92.90	94.71
Hepatitis	95.04	83.25	82.10	82.75	82.75	83.75
Ionosphere	95.04	82.02	83.17	83.97	82.06	84.71
Lungcancer	89.15	65.80	69.00	62.42	60.92	69.19
Lymphography	91.62	73.97	73.70	74.13	74.16	75.43
Spect	94.77	78.16	78.48	75.83	77.24	79.07
Tic-tac-toe	95.03	75.62	79.53	78.72	81.73	82.48
Vote	95.03	95.37	95.41	96.32	95.74	95.89
Wine	97.53	72.42	67.64	70.25	71.76	74.24
Zoo	97.14	94.14	89.35	90.55	88.91	90.89

表5 FSNB与5种基于粗糙集的优化特征选择算法的所选特征数实验结果

Tab. 5 Experimental results of FSNB and five optimized feature selection algorithms based on rough set on number of selected features

数据集	FSNB	ARRSFA	PSORSFS	FSARSR	QCSIA_FS	DQBGOA_MR
Average	8.49	7.23	9.02	8.65	7.34	5.92
Breastcancer	5.75	3.70	3.80	4.00	3.50	3.00
Hepatitis	14.65	4.10	6.20	5.60	5.00	3.70
Ionosphere	14.65	4.60	10.20	8.50	6.10	4.10
Lungcancer	6.30	7.90	19.40	17.10	11.60	4.30
Lymphography	14.75	9.50	8.60	7.80	7.80	7.10
Spect	5.00	14.00	15.10	16.10	14.70	14.00
Tic-tac-toe	5.00	7.20	7.40	8.00	7.00	7.00
Vote	5.00	9.50	8.80	9.60	8.10	7.50
Wine	7.20	3.20	3.60	3.60	3.10	3.00
Zoo	6.55	8.60	7.10	6.20	6.50	5.50

表6 FSNB与8种结合粗糙集与优化算法的特征选择算法的分类精度实验结果 ( %)

Tab. 6 Experimental results of FSNB and eight feature selection algorithms combining rough set and optimization algorithm on classification accuracy

数据集	FSNB	GARS	PSORS	ABCRS	FARS	SSORS	CSRS	HSRS	IRRARS
Average	89.70	76.49	78.72	82.57	79.81	83.22	82.59	82.48	85.79
Breastcancer	95.67	87.28	91.05	96.49	92.73	95.39	93.42	97.37	97.99
Congress	77.02	89.66	90.41	91.38	91.38	95.86	93.79	92.79	95.17
Heart	76.98	65.06	66.41	70.15	69.41	71.57	70.15	73.13	75.18
Ionosphere	95.04	79.85	80.40	84.90	80.80	84.70	85.40	84.00	86.07
Lungcancer	89.15	60.00	65.36	69.64	64.60	69.64	70.64	66.00	75.98
Waveform	96.89	74.82	76.82	80.60	77.90	81.90	80.60	80.20	83.25
Zoo	97.14	78.79	80.61	84.85	81.88	83.50	84.12	83.88	86.88

表7 FSNB与8种结合粗糙集与优化算法的特征选择算法所选特征数实验结果

Tab. 7 Experimental results of FSNB and eight feature selection algorithms combining rough set and optimization algorithm on number of selected features

数据集	FSNB	GARS	PSORS	ABCRS	FARS	SSORS	CSRS	HSRS	IRRARS
Average	7.68	710.79	602.76	635.32	619.06	596.46	665.62	585.41	376.36
Breastcancer	5.75	492.10	399.13	399.13	437.57	336.92	399.13	422.20	329.23
Congress	5.05	254.91	219.24	146.16	140.51	151.38	137.90	143.12	135.29
Heart	9.15	140.94	120.69	117.99	120.15	112.59	124.20	116.10	113.67
Ionosphere	14.65	193.75	188.49	133.03	144.96	143.21	132.68	143.21	134.08
Lungcancer	6.30	17.15	13.92	13.86	12.48	11.62	11.52	11.74	11.52
Waveform	6.30	3 815.00	3 225.00	3 600.00	3 435.00	3 385.00	3 805.00	3 220.00	1 880.00
Zoo	6.55	61.71	52.82	37.07	42.72	34.54	48.88	41.51	30.70

表8 FSNB与FSNM在分类精度、适应度值和所选特征数上的实验结果

Tab. 8 Experimental results of FSNM and FSNB on classification accuracy， fitness value and number of selected features

数据集	分类精度/%		适应度值		所选特征数		运行时间/s
数据集	FSNM	FSNB	FSNM	FSNB	FSNM	FSNB	FSNM	FSNB
Average	87.47	91.51	0.274 8	0.231 3	10.45	8.64	251	198
Breastcancer	95.24	95.67	0.370 8	0.342 8	6.25	5.75	119	118
Congress	72.68	77.02	0.208 9	0.209 0	5.00	5.05	55	55
Heart	75.22	76.98	0.401 9	0.408 3	8.15	9.15	33	33
Hepatitis	94.64	95.04	0.277 5	0.272 0	15.40	14.65	1 075	1 073
Ionosphere	94.10	95.04	0.276 1	0.272 0	15.10	14.65	1 074	1 065
Lungcancer	73.77	89.15	0.138 6	0.105 0	11.00	6.30	6	5
Lymphography	84.09	91.62	0.280 7	0.085 1	25.22	14.75	865	45
Sonar	81.87	84.21	0.200 2	0.173 6	17.80	15.55	24	24
Spect	86.96	94.77	0.297 8	0.220 5	6.00	5.00	52	81
Tic-tac-toe	85.62	95.03	0.313 9	0.219 5	6.20	5.00	80	143
Vote	93.51	95.03	0.303 6	0.219 5	8.60	5.00	80	85
Waveform	95.02	96.89	0.236 7	0.221 4	6.50	6.30	18	17
Wine	96.14	97.53	0.300 8	0.261 5	7.95	7.20	21	21
Zoo	95.72	97.14	0.240 2	0.227 9	7.10	6.55	18	16

表9 FSNB与8种结合邻域粗糙集与优化算法的特征选择算法的分类精度实验结果 ( %)

Tab. 9 Experimental results of FSNB and eight feature selection algorithms combining NRS and optimization algorithms on classification accuracy

数据集	FSNB	GANRS	PSONRS	ABCNRS	FANRS	SSONRS	CSNRS	HSNRS	IRRANRS
Average	89.70	79.24	83.34	85.92	81.44	86.23	85.42	86.16	88.57
Breastcancer	95.67	95.18	96.93	97.81	93.42	96.49	95.61	97.37	98.68
Congress	77.02	90.43	97.24	95.86	94.48	96.93	94.48	95.17	96.93
Heart	76.98	66.52	69.10	71.54	70.86	72.91	71.61	75.00	76.80
Ionosphere	95.04	80.58	81.39	86.02	81.98	85.60	86.27	85.49	86.61
Lungcancer	89.15	62.86	68.60	70.22	65.37	71.56	72.59	69.26	79.26
Waveform	96.89	78.92	80.88	83.04	79.42	83.18	82.96	84.99	84.64
Zoo	97.14	80.21	89.22	96.97	84.58	96.97	94.44	95.87	97.06

表10 FSNB与8种结合邻域粗糙集与优化算法的特征选择算法所选特征数实验结果

Tab. 10 Experimental results of FSNB and eight feature selection algorithms combining NRS and optimization algorithms on number of selected features

数据集	FSNB	GANRS	PSONRS	ABCNRS	FANRS	SSONRS	CSNRS	HSNRS	IRRANRS
Average	7.68	685.39	576.52	608.29	583.31	579.51	644.16	563.34	354.53
Breastcancer	5.75	466.23	381.79	359.99	370.82	355.51	389.90	399.55	312.87
Congress	5.05	245.21	206.58	136.81	128.19	140.33	124.85	134.07	124.67
Heart	9.15	124.71	119.53	110.70	105.30	108.86	122.15	112.48	106.22
Ionosphere	14.65	183.68	179.43	124.08	137.45	133.20	123.48	134.96	125.27
Lungcancer	6.30	16.39	13.10	13.19	11.82	10.81	10.63	10.81	11.09
Waveform	6.30	3 702.00	3 085.00	3 478.50	3 289.50	3 276.00	3 692.00	3 112.50	1 773.00
Zoo	6.55	59.52	50.20	34.78	40.05	31.83	46.11	39.04	28.60

表11 FSNB与6种基于二进制灰狼优化的特征选择算法的分类精度实验结果 ( %)

Tab. 11 Experimental results of FSNB and six feature selection algorithms based on binary grey wolf optimizer on classification accuracy

数据集	FSNB	BGWO	ABGWO	ABGWO-V1	ABGWO-V2	ABGWO-V3	ABGWO-V4
Average	91.93	77.61	78.39	78.29	78.04	78.08	78.14
Breastcancer	95.67	80.48	80.58	80.42	80.24	80.59	79.96
Congress	77.02	83.45	84.02	83.32	83.18	83.19	83.57
Ionosphere	95.04	85.95	86.90	86.90	86.91	86.63	86.37
Lymphography	91.62	46.24	47.27	47.27	47.33	47.14	47.55
Sonar	84.21	80.22	81.23	81.23	81.73	81.57	81.17
Spect	94.77	73.31	74.69	74.69	73.95	74.06	73.71
Tic-tac-toe	95.03	77.46	76.15	76.15	75.66	75.75	76.71
Waveform	96.89	81.71	82.38	82.38	81.56	81.74	81.79
Zoo	97.14	89.69	92.28	92.28	91.81	92.09	92.39

表12 FSNB与6种基于二进制灰狼优化的特征选择算法所选特征数实验结果

Tab. 12 Experimental results of FSNB and six feature selection algorithms based on binary grey wolf optimizer on number of selected features

数据集	FSNB	BGWO	ABGWO	ABGWO-V1	ABGWO-V2	ABGWO-V3	ABGWO-V4
Average	8.73	16.62	13.19	12.78	12.53	12.14	12.86
Breastcancer	5.75	5.30	3.80	4.45	3.75	3.95	4.05
Congress	5.05	9.90	8.25	7.75	8.10	7.40	8.50
Ionosphere	14.65	21.00	16.75	15.10	14.00	14.90	15.75
Lymphography	14.75	11.85	8.85	8.65	8.00	7.65	9.00
Sonar	15.55	36.60	30.45	28.20	28.65	27.25	27.85
Spect	5.00	15.45	11.15	11.65	11.85	10.65	11.60
Tic-tac-toe	5.00	6.80	5.90	6.00	5.30	5.35	5.85
Waveform	6.30	31.30	24.35	23.75	24.55	23.05	24.10
Zoo	6.55	11.40	9.20	9.45	8.60	9.10	9.05

表13 FSNB与5种基于粗糙集的优化特征选择算法的统计检验

Tab. 13 Statistical test of FSNB and five optimized feature selection algorithms based on rough set

平均排序						$x F 2$	$F F$
FSNB	ARRSFA	PSORSFS	FSARSR	QCSIA_FS	DQBGOA_MR	$x F 2$	$F F$
1.50	4.20	4.60	4.35	4.25	2.10	25.56	9.41

表13 FSNB与5种基于粗糙集的优化特征选择算法的统计检验

Tab. 13 Statistical test of FSNB and five optimized feature selection algorithms based on rough set

平均排序						$x F 2$	$F F$
FSNB	ARRSFA	PSORSFS	FSARSR	QCSIA_FS	DQBGOA_MR	$x F 2$	$F F$
1.50	4.20	4.60	4.35	4.25	2.10	25.56	9.41

表14 FSNB与8种结合粗糙集与优化算法的特征选择算法的统计检验

Tab. 14 Statistical test of FSNB and eight feature selection algorithms combining rough set and optimization algorithms

平均排序									$x F 2$	$F F$
FSNB	GARS	PSORS	ABCRS	FARS	SSORS	CSRS	HSRS	IRRARS	$x F 2$	$F F$
2.57	8.86	7.71	4.29	6.93	4.07	4.14	4.57	1.86	41.09	16.54

表14 FSNB与8种结合粗糙集与优化算法的特征选择算法的统计检验

Tab. 14 Statistical test of FSNB and eight feature selection algorithms combining rough set and optimization algorithms

平均排序									$x F 2$	$F F$
FSNB	GARS	PSORS	ABCRS	FARS	SSORS	CSRS	HSRS	IRRARS	$x F 2$	$F F$
2.57	8.86	7.71	4.29	6.93	4.07	4.14	4.57	1.86	41.09	16.54

表15 FSNB与8种结合邻域粗糙集与优化算法的特征选择算法的统计检验

Tab. 15 Statistical test of FSNB and eight feature selection algorithms combining NRS and optimization algorithms

平均排序									$x F 2$	$F F$
FSNB	GANRS	PSONRS	ABCNRS	FANRS	SSONRS	CSNRS	HSNRS	IRRANRS	$x F 2$	$F F$
2.86	8.71	6.00	4.21	7.64	3.93	5.29	4.29	2.07	34.77	9.83

表15 FSNB与8种结合邻域粗糙集与优化算法的特征选择算法的统计检验

Tab. 15 Statistical test of FSNB and eight feature selection algorithms combining NRS and optimization algorithms

平均排序									$x F 2$	$F F$
FSNB	GANRS	PSONRS	ABCNRS	FANRS	SSONRS	CSNRS	HSNRS	IRRANRS	$x F 2$	$F F$
2.86	8.71	6.00	4.21	7.64	3.93	5.29	4.29	2.07	34.77	9.83

表16 FSNB与6种基于二进制灰狼优化的特征选择算法的统计检验

Tab. 16 Statistical test of FSNB and six feature selection algorithms based on binary grey wolf optimizer

平均排序							$x F 2$	$F F$
FSNB	BGWO	ABGWO	ABGWO-V1	ABGWO-V2	ABGWO-V3	ABGWO-V4	$x F 2$	$F F$
1.67	5.56	3.28	3.83	4.78	4.67	4.22	18.51	4.17

表16 FSNB与6种基于二进制灰狼优化的特征选择算法的统计检验

Tab. 16 Statistical test of FSNB and six feature selection algorithms based on binary grey wolf optimizer

平均排序							$x F 2$	$F F$
FSNB	BGWO	ABGWO	ABGWO-V1	ABGWO-V2	ABGWO-V3	ABGWO-V4	$x F 2$	$F F$
1.67	5.56	3.28	3.83	4.78	4.67	4.22	18.51	4.17

图1 FSNB与5种基于粗糙集的优化特征选择算法的Bonferroni-Dunn检验结果

Fig. 1 Bonferroni-Dunn test results of FSNB and five optimized feature selection algorithms based on rough set

图2 FSNB与8种结合粗糙集与优化算法的特征选择算法的Bonferroni-Dunn检验结果

Fig. 2 Bonferroni-Dunn test results of FSNB and eight feature selection algorithms based on rough set and optimization algorithms

图3 FSNB与8种结合邻域粗糙集与优化算法的特征选择算法的Bonferroni-Dunn检验结果

Fig. 3 Bonferroni-Dunn test results of FSNB and eight feature selection algorithms based on NRS and optimization algorithms

图4 FSNB与6种基于二进制灰狼优化的特征选择算法的Bonferroni-Dunn检验结果

Fig. 4 Bonferroni-Dunn test results of FSNB and six feature selection algorithms based on binary grey wolf optimizer

参考文献 35

1	吴信东，董丙冰，堵新政，等.数据治理技术［J］.软件学报，2019，30（9）：2830-2856. 10.13328/j.cnki.jos.005854
	WU X D， DONG B B， DU X Z， et al. Data governance technology ［J］. Journal of Software， 2019， 30（9）： 2830-2856. 10.13328/j.cnki.jos.005854
2	邓威，郭钇秀，李勇，等.基于特征选择和Stacking集成学习的配电网网损预测［J］.电力系统保护与控制，2020，48（15）：108-115. 10.19783/j.cnki.pspc.191097
	DENG W， GUO Y X， LI Y， et al. Power losses prediction based on feature selection and Stacking integrated learning ［J］. Power System Protection and Control， 2020， 48（15）： 108-115. 10.19783/j.cnki.pspc.191097
3	HU P， PAN J S， CHU S C. Improved binary grey wolf optimizer and its application for feature selection ［J］. Knowledge-Based Systems， 2020， 195： Article No.105746. 10.1016/j.knosys.2020.105746
4	TUBISHAT M， IDRIS N， SHUIB L， et al. Improved salp swarm algorithm based on opposition based learning and novel local search algorithm for feature selection ［J］. Expert Systems with Applications， 2020， 145： Article No.113122. 10.1016/j.eswa.2019.113122
5	刘艳，程璐，孙林.基于K-S检验和邻域粗糙集的特征选择方法［J］.河南师范大学学报（自然科学版），2019，47（2）：21-28. 10.16366/j.cnki.1000-2367.2019.02.004
	LIU Y， CHENG L， SUN L. Feature selection method based on K-S test and neighborhood rough sets ［J］. Journal of Henan Normal University （Natural Science Edition）， 2019， 47（2）： 21-28. 10.16366/j.cnki.1000-2367.2019.02.004
6	刘琨，封硕.加强局部搜索能力的人工蜂群算法［J］.河南师范大学学报（自然科学版），2021，49（2）：15-24. 10.16366/j.cnki.1000-2367.2021.02.003
	LIU K， FENG S. An improved artificial bee colony algorithm for enhancing local search ability ［J］. Journal of Henan Normal University （Natural Science Edition）， 2021， 49（2）： 15-24. 10.16366/j.cnki.1000-2367.2021.02.003
7	ZAIMOGLU E A， CELEBI N， YURTAY N. Binary-coded tug of war optimization algorithm for attribute reduction based on rough set ［J］. Journal of Multiple-Valued Logic and Soft Computing， 2020， 35（1/2）： 93-111. 10.1007/978-3-030-23756-1_144
8	HUDA R K， BANKA H. Efficient feature selection and classification algorithm based on PSO and rough sets ［J］. Neural Computing and Applications， 2019， 31（8）： 4287-4303. 10.1007/s00521-017-3317-9
9	TAWHID M A， IBRAHIM A M. Feature selection based on rough set approach， wrapper approach， and binary whale optimization algorithm ［J］. International Journal of Machine Learning and Cybernetics， 2020， 11（3）： 573-602. 10.1007/s13042-019-00996-5
10	WANG D， CHEN H M， LI T R， et al. A novel quantum grasshopper optimization algorithm for feature selection ［J］. International Journal of Approximate Reasoning， 2020， 127： 33-53. 10.1016/j.ijar.2020.08.010
11	AZIZ M ABD EL， HASSANIEN A E. Modified cuckoo search algorithm with rough sets for feature selection ［J］. Neural Computing and Applications， 2018， 29（4）： 925-934. 10.1007/s00521-016-2473-7
12	ZOU L， LI H X， JIANG W， et al. An improved fish swarm algorithm for neighborhood rough set reduction and its application ［J］. IEEE Access， 2019， 7： 90277-90288. 10.1109/access.2019.2926799
13	薛占熬，庞文莉，姚守倩，等.基于前景理论的直觉模糊三支决策模型［J］.河南师范大学学报（自然科学版），2020，48（5）：31-36，79.
	XUE Z A， PANG W L， YAO S Q， et al. The prospect theory based intuitionistic fuzzy three-way decisions model ［J］. Journal of Henan Normal University （Natural Science Edition）， 2020， 48（5）： 31-36， 79.
14	HU Q H， YU D R， LIU J F， et al. Neighborhood rough set based heterogeneous feature subset selection ［J］. Information Sciences， 2018， 178（18）： 3577-3594.
15	彭鹏，倪志伟，朱旭辉，等.基于改进二元萤火虫群优化算法和邻域粗糙集的属性约简方法［J］.模式识别与人工智能，2020，33（2）：95-105.
	PENG P， NI Z W， ZHU X H， et al. Attribute reduction method based on improved binary glowworm swarm optimization algorithm and neighborhood rough set ［J］. Pattern Recognition and Artificial Intelligence， 2020， 33（2）： 95-105.
16	SUN L， CHEN S S， XU J C， et al. Improved monarch butterfly optimization algorithm based on opposition-based learning and random local perturbation ［J］. Complexity， 2019， 2019： Article No.4182148. 10.1155/2019/4182148
17	WANG G G， DEB S， CUI Z H. Monarch butterfly optimization ［J］. Neural Computing and Applications， 2019， 31（7）： 1995-2014. 10.1007/s00521-015-1923-y
18	DORGHAM O M， ALWESHAH M， RYALAT M H， et al. Monarch butterfly optimization algorithm for computed tomography image segmentation ［J］. Multimedia Tools and Applications， 2021， 80（20）： 30057-30090. 10.1007/s11042-020-10147-6
19	SOLTANI P， HADAVANDI E. A monarch butterfly optimization- based neural network simulator for prediction of siro-spun yarn tenacity ［J］. Soft Computing， 2019， 23（20）： 10521-10535. 10.1007/s00500-018-3624-9
20	孙林，赵婧，徐久成，等.基于改进帝王蝶优化算法的特征选择方法［J］.模式识别与人工智能，2020，33（11）：981-994.
	SUN L， ZHAO J， XU J C， et al. Feature selection method based on improved monarch butterfly optimization algorithm ［J］. Pattern Recognition and Artificial Intelligence， 2020， 33（11）： 981-994.
21	MAFARJA M， ALJARAH I， FARIS H， et al. Binary grasshopper optimisation algorithm approaches for feature selection problems ［J］. Expert Systems with Applications， 2019， 117： 267-286. 10.1016/j.eswa.2018.09.015
22	SUN L， WANG L Y， DING W P， et al. Feature selection using fuzzy neighborhood entropy-based uncertainty measures for fuzzy neighborhood multigranulation rough sets ［J］. IEEE Transactions on Fuzzy Systems， 2021， 29（1）： 19-33. 10.1109/tfuzz.2020.2989098
23	LONG N C， MEESAD P， UNGER H. Attribute reduction based on rough sets and the discrete firefly algorithm ［M］// BOONKRONG S， UNGER H， MEESAD P. Recent Advances in Information and Communication Technology， AISC 265. Cham： Springer， 2014： 13-22.
24	WANG X Y， YANG J， TENG X L， et al. Feature selection based on rough sets and particle swarm optimization ［J］. Pattern Recognition Letters， 2007， 28（4）： 459-471. 10.1016/j.patrec.2006.09.003
25	CHEN Y M， ZHU Q X， XU H R. Finding rough set reducts with fish swarm algorithm ［J］. Knowledge-Based Systems， 2015， 81： 22-29. 10.1016/j.knosys.2015.02.002
26	ZOUACHE D， ABDELAZIZ F BEN. A cooperative swarm intelligence algorithm based on quantum-inspired and rough sets for feature selection ［J］. Computers and Industrial Engineering， 2018， 115： 26-36. 10.1016/j.cie.2017.10.025
27	IBRAHIM R A， ELAZIZ M ABD， OLIVA D， et al. An improved runner-root algorithm for solving feature selection problems based on rough sets and neighborhood rough sets ［J］. Applied Soft Computing， 2020， 97（Pt B）： Article No.105517. 10.1016/j.asoc.2019.105517
28	JING S Y. A hybrid genetic algorithm for feature subset selection in rough set theory ［J］. Soft Computing， 2014， 18（7）： 1373-1382. 10.1007/s00500-013-1150-3
29	ZHANG Y， GONG D W， HU Y， et al. Feature selection algorithm based on bare bones particle swarm optimization ［J］. Neurocomputing， 2015， 148： 150-157. 10.1016/j.neucom.2012.09.049
30	SUGUNA N， THANUSHKODI K G. An independent rough set approach hybrid with artificial bee colony algorithm for dimensionality reduction ［J］. American Journal of Applied Sciences， 2011， 8（3）： 261- 266. 10.3844/ajassp.2011.261.266
31	汪春峰，褚新月.基于精英邻居引导的萤火虫算法［J］.河南师范大学学报（自然科学版），2019，47（6）：15-21.
	WAND C F， CHU X Y. A firefly algorithm based on elite neighborhood guide ［J］. Journal of Henan Normal University （Natural Science Edition）， 2019， 47（6）： 15-21.
32	AZIZ M ABD EL， HASSANIEN A E. An improved social spider optimization algorithm based on rough sets for solving minimum number attribute reduction problem ［J］. Neural Computing and Applications， 2018， 30（8）： 2441-2452. 10.1007/s00521-016-2804-8
33	徐小琴，王博，赵红生，等.基于布谷鸟搜索和模拟退火算法的两电压等级配网重构方法［J］.电力系统保护与控制，2020，48（11）：84-91.
	XU X Q， WANG B， ZHAO H S， et al. Reconfiguration of two-voltage distribution network based on cuckoo search and simulated annealing algorithm ［J］. Power System Protection and Control， 2020， 48（15）： 108-115.
34	INBARANI H H， BAGYAMATHI M， AZAR A T. A novel hybrid feature selection method based on rough set and improved harmony search ［J］. Neural Computing and Applications， 2015， 26（8）： 1859-1880. 10.1007/s00521-015-1840-0
35	SUN L， WANG L Y， DING W P， et al. Neighborhood multi- granulation rough sets-based attribute reduction using Lebesgue and entropy measures in incomplete neighborhood decision systems ［J］. Knowledge-Based Systems， 2020， 192： Article No.105373. 10.1016/j.knosys.2019.105373

[1]	李晓寒, 贾华丁, 程雪, 李太勇. 基于改进遗传算法和图神经网络的股市波动预测方法[J]. 《计算机应用》唯一官方网站, 2022, 42(5): 1624-1633.
[2]	李莉, 石可欣, 任振康. 基于特征选择和TrAdaBoost的跨项目缺陷预测方法[J]. 《计算机应用》唯一官方网站, 2022, 42(5): 1554-1562.
[3]	夏冰, 庞建民, 周鑫, 单征. 二进制代码相似性搜索研究进展[J]. 《计算机应用》唯一官方网站, 2022, 42(4): 985-998.
[4]	轩书婷, 刘惊雷. 基于离散哈希的聚类[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 713-723.
[5]	张小清, 王晨曦, 吕彦, 林耀进. 基于ReliefF的层次分类在线流特征选择算法[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 688-694.
[6]	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.
[7]	陈永波, 李巧勤, 刘勇国. 基于动态相关性的特征选择算法[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 109-114.
[8]	湛航, 何朗, 黄樟灿, 李华峰, 张蔷, 谈庆. 改进的基于层次距离的基因表达式编程特征选择分类算法[J]. 计算机应用, 2021, 41(9): 2658-2667.
[9]	祝承, 赵晓琦, 赵丽萍, 焦玉宏, 朱亚飞, 陈建英, 周伟, 谭颖. 基于谱聚类半监督特征选择的功能磁共振成像数据分类[J]. 计算机应用, 2021, 41(8): 2288-2293.
[10]	李蒙蒙, 秦伟, 刘艺, 刁兴春. 结合头脑风暴优化的混合蚁群优化算法[J]. 计算机应用, 2021, 41(8): 2412-2417.
[11]	刘双元, 郑王里, 林云汉. 面向三维特征描述子的自适应二进制简化方法[J]. 计算机应用, 2021, 41(7): 2062-2069.
[12]	林筠超, 万源. 基于图结构优化的自适应多度量非监督特征选择方法[J]. 计算机应用, 2021, 41(5): 1282-1289.
[13]	贾鹤鸣, 姜子超, 李瑶, 孙康健. 基于改进斑点鬣狗优化算法的同步优化特征选择[J]. 计算机应用, 2021, 41(5): 1290-1298.
[14]	王泽昆, 贺毅朝, 李焕哲, 张发展. 基于新颖S型转换函数的二进制粒子群优化算法求解具有单连续变量的背包问题[J]. 计算机应用, 2021, 41(2): 461-469.
[15]	杨柳, 李云. 混合式的K-匿名特征选择算法[J]. 《计算机应用》唯一官方网站, 2021, 41(12): 3521-3526.

基于邻域粗糙集和帝王蝶优化的特征选择算法

Feature selection algorithm based on neighborhood rough set and monarch butterfly optimization

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 20

参考文献 35

相关文章 15

编辑推荐

Metrics