基于天牛群优化与改进正则化极限学习机的网络入侵检测

王振东; 刘尧迪; 杨书新; 王俊岭; 李大海

doi:10.16383/j.aas.c190851

基于天牛群优化与改进正则化极限学习机的网络入侵检测

doi: 10.16383/j.aas.c190851

1.
江西理工大学信息工程学院赣州 341000

基金项目: 国家自然科学基金(61562037, 61562038, 61563019, 61763017), 江西省自然科学基金(20171BAB202026, 20181BBE58018)资助

详细信息

作者简介:
王振东：博士, 江西理工大学信息工程学院副教授. 主要研究方向为无线传感器网络, 智能物联网, 认知计算, 大数据和信息安全.E-mail: wangzhendong@hrbeu.edu.cn

刘尧迪：江西理工大学信息工程学院硕士研究生. 主要研究方向为网络安全, 入侵检测, 群智能优化算法, 机器学习与深度学习. 本文通信作者.E-mail: liuyaodi@yeah.net

杨书新：博士, 江西理工大学信息工程学院副教授. 主要研究方向为数据管理, 信息检索和生物信息学.E-mail: yimuyunlang@sina.com

王俊岭：博士, 江西理工大学信息工程学院副教授. 主要研究方向为分布式计算, 容错, 计算机视觉.E-mail: wangjunling@jxust.edu.cn

李大海：博士, 江西理工大学信息工程学院副教授. 主要研究方向为分布式系统服务质量控制, 分布式系统自学资源调度控制.E-mail: dlai6535@aliyun.com

计量
- 文章访问数: 691
- HTML全文浏览量: 266
- PDF下载量: 160
- 被引次数: 0
出版历程
- 收稿日期: 2019-12-16
- 录用日期: 2020-04-07
- 网络出版日期: 2022-11-29
- 刊出日期: 2022-12-23

Network Intrusion Detection Based BSO and Improved RELM

1.
School of Information Engineering, Jiangxi University of Science and Technology, Ganzhou 341000

Funds: Supported by National Natural Science Foundation of China (61562037, 61562038, 61563019, 61763017) and Natural Science Foundation of Jiangxi Province (20171BAB202026, 20181BBE58018)

More Information

Author Bio:
WANG Zhen-Dong　Ph.D., associate professor at the School of Information Engineering, Jiangxi University of Science and Technology. His research interest covers wireless sensor networks, smart internet of things, cognitive computing, and big data and information security

LIU Yao-Di　Master student at the School of Information Engineering, Jiangxi University of Science and Technology. Her research interest covers network security, intrusion detection, swarm intelligence optimization algorithm, machine learning, and deep learning. Corresponding author of this paper

YANG Shu-Xin　Ph.D., associate professor at the School of Information Engineering, Jiangxi University of Science and Technology. His research interest covers data management, information retrieval, and bioinformatics

WANG Jun-Ling　Ph.D., associate professor at the School of Information Engineering, Jiangxi University of Science and Technology. His research interest covers distributed computing, fault tolerance, and computer vision

LI Da-Hai　Ph.D., associate professor at the School of Information Engineering, Jiangxi University of Science and Technology. His research interest covers distributed system quality of service (QoS) control, and distributed system self-learning resource scheduling control

摘要

摘要: 正则化极限学习机(Regularized extreme learning machine, RELM)因其极易于实现、训练速度快等优点在诸多领域均取得了成功应用. 对此, 本文将RELM引入到入侵检测中, 设计了天牛群优化算法(Beetle swarm optimization, BSO), 并针对RELM由于随机初始化参数带来的潜在缺陷, 提出基于天牛群优化与改进正则化极限学习机(BSO-IRELM)的网络入侵检测算法. 使用LU分解求解RELM的输出权值矩阵, 进一步缩短了RELM的训练时间, 同时利用BSO对RELM的权值和阈值进行联合优化. 为避免BSO算法陷入局部最优, 引入Tent映射反向学习、莱维飞行的群体学习与动态变异策略提升优化性能. 实验结果表明, 在机器学习UCI数据集上, 相比于RELM、IRELM、GA-IRELM、PSO-IRELM等算法, BSO-IRELM的数据分类性能提升明显. 最后, 将BSO-IRELM应用于网络入侵检测数据集NSL-KDD, 并与BP (Back propagation)、LR (Logistics regression)、RBF (Radial basis function)、AB (AdaBoost)、SVM (Support vector machine)、RELM、IRELM等算法进行了对比, 结果证明BSO-IRELM算法在准确率、精确率、真正率和假正率等指标上均具有明显优势.
- 入侵检测 /
- 正则化极限学习机 /
- LU分解 /
- 天牛群优化算法
Abstract: Regularized extreme learning machine (RELM) has been successfully applied in many fields due to its advantages of easy realization and fast training speed. In this regard, RELM is introduced into intrusion detection. Potential flaws caused by random initialization parameters, and a network intrusion detection algorithm based on the beetle swarm optimization (BSO) and improved RELM (BSO-IRELM) was proposed. LU decomposition was used to solve the output of weighting matrix of RELM, which further shortened the training of RELM. At the same time, BSO algorithm was designed to jointly optimize the weights and thresholds of RELM. To prevent the BSO algorithm from falling into a local optimum, Tent mapping inverse learning, Levy flight group learning and dynamic mutation strategies were introduced to improve optimization performance. Experimental results show that compared to algorithms such as RELM, IRELM, GA-IRELM, and PSO-IRELM on machine learning UCI datasets, BSO-IRELM's data classification performance improves significantly. Finally, the BSO-IRELM was applied to the network intrusion detection dataset NSL-KDD, and compared with BP (back propagation), LR (logistics regression), RBF (radial basis function), AB (AdaBoost), SVM (support vector machine), RELM, IRELM and other algorithms. It is clear that the BSO-IRELM algorithm has obvious advantages in terms of accuracy, precision, true positive rate, and false positive rate.
- Intrusion detection /
- regularized extreme learning machine (RELM) /
- LU decomposition /
- beetle swarm optimization (BSO)

HTML全文

图 1 莱维飞行轨迹示意图

Fig. 1 Levy flight path diagram

下载: 全尺寸图片幻灯片

图 2 柯西变异概率分布图

Fig. 2 Cauchy variation probability distribution map

下载: 全尺寸图片幻灯片

图 3 天牛群算法流程图

Fig. 3 Flow chart of BSO algorithm

下载: 全尺寸图片幻灯片

图 4 算法测试结果图

Fig. 4 Test result graph of algorithm

下载: 全尺寸图片幻灯片

图 5 BSO-IRELM算法入侵检测框架图

Fig. 5 BSO-IRELM Algorithm intrusion detection framework

下载: 全尺寸图片幻灯片

图 6 部分算法在UCI数据集上的检测结果

Fig. 6 The detection results of part algorithm on UCI dataset

下载: 全尺寸图片幻灯片

图 7 2元分类混淆矩阵

Fig. 7 Binary classification confusion matrix

下载: 全尺寸图片幻灯片

图 8 2元分类ROC曲线对比图

Fig. 8 Binary classification of ROC curve comparison diagram

下载: 全尺寸图片幻灯片

图 9 多元分类混淆矩阵

Fig. 9 Multiple classification confusion matrix

下载: 全尺寸图片幻灯片

图 10 多元分类ROC曲线对比图

Fig. 10 Multiple classification ROC curve comparison diagram

下载: 全尺寸图片幻灯片

表 1 群智能算法参数

Table 1 Swarm intelligence algorithm parameters

参数	值	参数	值
自学习因子	2.4	群体学习因子	1.6
交叉概率	0.7	遗传概率	0.5
惯性权重	0.8	种群数量	50
比例系数	0.4, 0.6	惩罚因子	0.5

下载: 导出CSV

表 2 UCI数据集

Table 2 UCI dataset

数据集	维度	类别数	样本总数	测试样本数
Iris	4	3	150	30
Wine	13	3	178	30

下载: 导出CSV

表 3 各算法在UCI数据集上的准确率(%)

Table 3 Accuracy of each algorithm on UCI dataset (%)

数据集	RELM	IRELM	GA-IRELM	PSO-IRELM	BSO-RELM	BSO-IRELM
Iris	76.6667 (23/30)	90 (27/30)	96.6667 (29/30)	100 (30/30)	100 (30/30)	100 (30/30)
Wine	80 (24/30)	90 (27/30)	93.3333 (28/30)	93.3333 (28/30)	93.3333 (28/30)	96.6666 (29/30)

下载: 导出CSV

表 4 各算法在Iris数据集上的性能评价指标

Table 4 Performance evaluation index of each algorithm on Iris dataset

算法	类别	精确率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
RELM	1	100 (9/9)	81.8182 (9/1)	0 (0/14)	90	0.9091
	2	50 (5/10)	83.3333 (5/6)	21.7391 (5/23)	62.5	0.8125
	3	81.8182 (9/11)	69.2308 (9/13)	12.5 (2/16)	75	0.7873
IRELM	1	100 (8/8)	100 (8/8)	0 (0/19)	100	1
	2	75 (9/12)	100 (9/9)	14.2857 (3/21)	85.7143	0.9286
	3	100 (10/10)	76.9231 (10/13)	0 (0/17)	86.9565	0.8846
GA-IRELM	1	100 (13/13)	100 (13/13)	0 (0/16)	100	1
	2	100 (7/7)	87.5 (7/8)	0 (0/22)	93.3333	0.9375
	3	90 (9/10)	100 (9/9)	4.7619 (1/21)	94.7368	0.9762
PSO-IRELM	1	100 (8/8)	100 (8/8)	0 (0/22)	100	1
	2	100 (12/12)	100 (12/12)	0 (0/28)	100	1
	3	100 (10/10)	100 (10/10)	0 (0/20)	100	1
BSO-RELM	1	100 (13/13)	100 (13/13)	0 (0/17)	100	1
	2	100 (10/10)	100 (10/10)	0 (0/20)	100	1
	3	100 (7/7)	100 (7/7)	0 (0/23)	100	1
BSO-IRELM	1	100 (12/12)	100 (12/12)	0 (0/18)	100	1
	2	100 (5/5)	100 (5/5)	0 (0/25)	100	1
	3	100 (13/13)	100 (13/13)	0 (0/17)	100	1

下载: 导出CSV

表 5 各算法在Wine数据集上的性能评价指标

Table 5 Performance evaluation index of each algorithm on Wine dataset

算法	类别	精确率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
RELM	1	81.8182 (9/11)	100 (9/9)	11. 7647 (2/17)	90	0.9524
	2	66.6667 (8/12)	80 (8/10)	20 (4/20)	72.7273	0.8000
	3	100 (7/7)	63.6364 (7/11)	0 (0/17)	77.7778	0.8182
IRELM	1	88.8889 (8/9)	88.8889 (8/9)	5 (1/20)	88.8889	0.9206
	2	90.9091 (10/11)	83.3333 (10/12)	5.5556 (1/18)	86.9565	0.8889
	3	90 (9/10)	100 (9/9)	5.2632 (1/19)	94.7368	0.9762
GA-IRELM	1	87.5 (7/8)	100 (7/7)	4. 5455 (1/22)	93. 3333	0.9783
	2	100 (16/16)	88. 8889 (16/18)	0 (0/12)	94.1176	0.9444
	3	83.3333 (5/6)	100 (5/5)	4.1667 (1/24)	90.9091	0.9800
PSO-IRELM	1	90 (9/10)	100 (9/9)	5 (1/20)	94.7368	0.9762
	2	90. 9091 (10/11)	90.9091 (10/11)	5.2632 (1/19)	90.9091	0.9282
	3	100 (9/9)	90 (9/10)	0 (0/19)	94.7368	0.9500
BSO-RELM	1	90.9091 (10/11)	100 (10/10)	5.2632 (1/19)	95.2381	0.9750
	2	87.5 (7/8)	87.5 (7/8)	4.5455 (1/22)	87.5	0.9148
	3	100 (11/11)	91.6667 (11/12)	0 (0/17)	95.6522	0.9583
BSO-IRELM	1	100 (12/12)	92.3077 (12/13)	0 (0/17)	96	0.9615
	2	92.3077 (12/13)	100 (12/12)	5.5556 (1/18)	96	0.9722
	3	100 (5/5)	100 (5/5)	0 (0/24)	100	1

下载: 导出CSV

表 6 各算法的准确率(%)

Table 6 Accuracy of each algorithm (%)

算法	准确率
BP	78.1 (1562/2000)
LR	81.3 (1626/2000)
RBF	88.9 (1778/2000)
AB	86.15 (1723/2000)
SVM	91.15 (1823/2000)
RELM	81.45 (1629/2000)
IRELM	83.9 (1678/2000)
GA-IRELM	89.5 (1790/2000)
PSO-IRELM	90.45 (1809/2000)
BSO-IRELM	91.25 (1825/2000)

下载: 导出CSV

表 7 各算法的性能评价指标

Table 7 Performance evaluation index of each algorithm

算法	类别	精确率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
BP	1	45.2915 (303/669)	80.8 (303/375)	22.5231 (366/1625)	58.046	0.7914
BP	2	94.5905 (1259/1331)	77.4769 (1259/1625)	19.2 (72/375)	85.1827	0.7914
LR	1	85.7143 (6/7)	1.5831 (6/379)	0.06169 (1/1621)	3.1088	0.5076
LR	2	81.2845 (1620/1993)	99.9383 (1620/1621)	98.4169 (373/379)	89.6514	0.5076
RBF	1	67.8663 (264/389)	73.1302 (264/361)	7.6266 (125/1639)	70.4	0.8275
RBF	2	93.9789 (1514/1611)	92.3734 (1514/1639)	26.8698 (97/361)	93.1692	0.8275
AB	1	67.9825 (155/228)	43.1755 (155/359)	4.4485 (73/1641)	52.8109	0.6936
AB	2	88.4876 (1568/1772)	95.5515 (1568/1641)	56.8245 (204/359)	91.884	0.6936
SVM	1	89.7674 (193/215)	55.4598 (193/348)	1.3317 (22/1652)	68.5613	0.7706
SVM	2	91.3165 (1630/1785)	98.6683 (1630/1652)	44.5402 (155/348)	94.8502	0.7706
RELM	1	53.6339 (140/261)	35.8974 (140/390)	7.5155 (121/1610)	43.0088	0.6711
RELM	2	85.6239 (1489/1739)	92.4844 (1489/1610)	64.1025 (250/390)	88.9220	0.7076
IRELM	1	55.5556 (145/261)	41.3105 (145/351)	7.0346 (116/1649)	47.3856	0.6714
IRELM	2	88.1541 (1533/1739)	92.9654 (1533/1649)	58.6894 (206/351)	90.4958	0.7280
GA-IRELM	1	88.8412 (207/233)	52.9412 (20/391)	1.6159 (26/1609)	66.3462	0.7566
GA-IRELM	2	89.5868 (1583/1767)	98.3840 (1583/1609)	47.0588 (184/391)	93.7792	0.7955
PSO-IRELM	1	84.5588 (230/272)	60.686 (230/379)	2.5910 (42/1621)	70.6605	0.7905
PSO-IRELM	2	91.3773 (1579/1728)	97.4090 (1579/1621)	39.3139 (149/379)	94.2967	0.8066
BSO-IRELM	1	86.747 (216/249)	60.3352 (216/358)	2.0097 (33/1642)	71.1697	0.7916
BSO-IRELM	2	91.9428 (1609/1751)	97.9902 (1609/1642)	39.6648 (142/358)	94.8702	0.8416

下载: 导出CSV

表 8 不同算法检测准确率(%)

Table 8 Accuracy of different algorithms (%)

算法	准确率
BP	73.1 (1462/2000)
LR	47.2 (944/2000)
RBF	81.95 (1639/2000)
AB	76.05 (1521/2000)
SVM	83.15 (1663/2000)
RELM	62.7 (1254/2000)
IRELM	71.9 (1438/2000)
GA-IRELM	86.35 (1727/2000)
PSO-IRELM	86.15 (1723/2000)
BSO-IRELM	88.7 (1774/2000)

下载: 导出CSV

表 9 各算法在Normal上的性能评价指标

Table 9 Performance evaluation index of each algorithm on Normal

算法	精准率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
BP	77.7778 (14/18)	3.8674 (14/362)	0.27548 (4/1452)	7.3684	0.5181
LR	76.4706 (13/17)	3.6723 (13/354)	0.42781 (4/935)	7.0081	0.5172
RBF	66.9377 (247/369)	73.5119 (247/336)	8.0581 (122/1514)	70.0709	0.8301
AB	52.3517 (256/489)	66.8407 (256/383)	15.5541 (233/1498)	58.7156	0.7622
SVM	91.2863 (220/241)	63.0372 (220/349)	1.4344 (21/1464)	74.5763	0.8088
RELM	89.5238 (188/210)	53.4091 (188/352)	2.0221 (22/1088)	66.9039	0.7604
IRELM	49.7619 (209/420)	58.3799 (209/358)	14.6528 (211/1440)	53.7275	0.7277
GA-IRELM	88.9706 (242/272)	64.191 (242/377)	1.9802 (30/1515)	74.5763	0.8117
PSO-IRELM	80.3571 (225/280)	61.3079 (225/367)	3.5415 (55/1553)	69.5518	0.7897
BSO-IRELM	83.9552 (225/268)	66.1765 (225/340)	2.701 (43/1592)	74.0132	0.8179

下载: 导出CSV

表 10 各算法在Probe上的性能评价指标

Table 10 Performance evaluation index of each algorithm on Probe

算法	精准率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
BP	82.9787 (390/470)	97.5 (390/400)	6.9444 (80/1152)	89.6552	0.9625
LR	62.6039 (226/361)	56.7839 (226/398)	15.8265 (135/853)	59.552	0.7418
RBF	99.5902 (243/244)	59.7052 (243/407)	0.071582 (1/1397)	74.6544	0.7982
AB	89.6359 (320/357)	82.4742 (320/388)	2.9887 (37/1238)	85.906	0.9009
SVM	73.3728 (372/507)	88.7828 (372/419)	9.467 (135/1426)	80.3456	0.9012
RELM	51.7661 (425/821)	97.7011 (425/435)	32.3265 (396/1225)	67.6752	0.8620
IRELM	79.9065 (342/428)	91.4439 (342/374)	7.2758 (86/1182)	85.2868	0.9308
GA-IRELM	89.8851 (391/435)	92.435 (391/423)	3.1884 (44/1380)	91.1422	0.9482
PSO-IRELM	89.7638 (3422/381)	94.4751 (342/362)	2.7465 (39/1420)	92.0592	0.9605
BSO-IRELM	89.6629 (399/445)	93.8824 (399/425)	3.2372 (396/1225)	91.7241	0.9548

下载: 导出CSV

表 11 各算法在DoS上的性能评价指标

Table 11 Performance evaluation index of each algorithm on DoS

算法	精准率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
BP	93.0233 (600/645)	81.5217 (600/736)	4.9614 (45/907)	86.8936	0.8898
LR	42.1687 (665/1577)	87.5 (665/760)	76.5743 (912/1191)	56.9106	0.5698
RBF	99.4074 (671/675)	90.9214 (671/738)	0.41152 (4/972)	94.9752	0.9530
AB	90.3509 (515/570)	70.6447 (515/729)	5.1838 (55/1061)	79.2918	0.8316
SVM	84.7118 (676/798)	90.1333 (676/750)	11.0009 (122/1109)	87.3385	0.9019
RELM	96.7391 (178/184)	25.0704 (178/710)	0.55453 (6/1082)	39.821	0.6230
IRELM	96.7059 (411/425)	54.0079 (411/761)	1.3449 (14/1041)	69.3086	0.7644
GA-IRELM	90.7539 (638/703)	89.4811 (638/713)	5.6326 (65/1154)	90.113	0.9222
PSO-IRELM	93.5302 (665/711)	89.502 (665/742)	4.1667 (16/1104)	91.4718	0.9292
BSO-IRELM	96.3636 (689/715)	93.6141 (689/736)	2.3402 (26/1111)	94.969	0.9578

下载: 导出CSV

表 12 各算法在R2L上的性能评价指标

Table 12 Performance evaluation index of each algorithm on R2L

算法	精准率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
BP	87.5 (14/16)	31.1111 (14/45)	0.13793 (2/1450)	45.9016	0.5150
LR	NaN (0/0)	0 (0/32)	0 (0/944)	NaN	0
RBF	68 (17/25)	56.6667 (17/30)	0.4908 (8/1630)	61.8182	0.7813
AB	NaN (0/0)	0 (0/36)	0 (0/1521)	NaN	0
SVM	NaN (0/0)	0 (0/36)	0 (0/1663)	NaN	0
RELM	50 (1/2)	3.4483 (1/29)	0.079745 (1/1254)	6.4516	0.5147
IRELM	81.8182 (9/11)	39.1304 (9/23)	0.13976 (2/1431)	52.9412	0.6951
GA-IRELM	90.9091 (20/22)	57.1429 (20/35)	0.11703 (2/1709)	70.1754	0.7852
PSO-IRELM	90.9091 (20/22)	54.0541 (20/37)	0.1173 (2/1705)	67.7966	0.7698
BSO-IRELM	92.3077 (24/26)	82.7586 (24/29)	0.39728 (7/1762)	69.0909	0.8258

下载: 导出CSV

表 13 各算法在U2R上的性能评价指标

Table 13 Performance evaluation index of each algorithm on U2R

算法	精准率 (%)	TPR (%)	FPR (%)	F值 (%)	AUC
BP	52.1739 (44/851)	97.1554 (444/457)	28.5614 (407/1425)	67.8899	0.8539
LR	88.8889 (40/45)	8.7719 (40/456)	0.55006 (5/909)	15.9681	0.5422
RBF	67.1033 (461/687)	94.274 (461/489)	16.0969 (226/1404)	78.4014	0.8966
AB	73.6301 (430/584)	92.6724 (430/464)	12.3695 (154/1245)	82.0611	0.9132
SVM	87.0044 (395/454)	88.565 (395/446)	4.4461 (59/1327)	87.7778	0.9238
RELM	59.0038 (462/783)	97.4684 (462/474)	28.841 (321/1113)	73.5084	0.8822
IRELM	65.2235 (467/716)	96.4876 (467/484)	20.4098 (249/1220)	77.8333	0.9003
GA-IRELM	76.7606 (436/568)	96.4602 (436/452)	9.2762 (132/1423)	85.4902	0.9397
PSO-IRELM	77.7228 (471/606)	95.9267 (471/491)	9.7332 (135/1387)	85.8706	0.9349
BSO-IRELM	80.9524 (442/546)	94.0426 (442/470)	7.2423 (104/1436)	87.0079	0.9362

下载: 导出CSV

参考文献(30)

[1]	Tsai C F, Hsu Y F, Lin C Y, Lin W Y. Intrusion detection by machine learning: A review. Expert Systems With Applications, 2009, 36(10): 11994-12000 doi: 10.1016/j.eswa.2009.05.029
[2]	任家东, 刘新倩, 王倩, 何海涛, 赵小林. 基于KNN离群点检测和随机森林的多层入侵检测方法. 计算机研究与发展, 2019, 56(3): 566-575 Ren Jia-Dong, Liu Xin-Qian, Wang Qian, He Hai-Tao, Zhao Xiao-Lin. An multi-level intrusion detection method based on KNN outlier detection and random forests. Journal of Computer Research and Development, 2019, 56(3): 566-575
[3]	高妮, 高岭, 贺毅岳, 王海. 基于自编码网络特征降维的轻量级入侵检测模型. 电子学报, 2017, 45(3): 730-739 Gao Ni, Gao Ling, HE Yi-Yue, Wang Hai. A lightweight intrusion detection model based on autoencoder network with feature reduction. Acta Electronica Sinica, 2017, 45(3): 730-739
[4]	Ahmad I, Basheri M, Iqbal M J, Rahim A. Performance comparison of support vector machine, random forest, and extreme learning machine for intrusion detection. IEEE Access, 2018, 6: 33789-33795 doi: 10.1109/ACCESS.2018.2841987
[5]	Mabu S, Gotoh S, Obayashi M, Kuremoto T. A random-forests-based classifier using class association rules and its application to an intrusion detection system. Artificial Life and Robotics, 2016, 21(3): 371-377 doi: 10.1007/s10015-016-0281-x
[6]	Shenfield A, Day D, Ayesh A. Intelligent intrusion detection systems using artificial neural networks. ICT Express, 2018, 4(2): 95-99 doi: 10.1016/j.icte.2018.04.003
[7]	Ding H W, Wan L. Research on intrusion detection based on KPCA-BP neural network. In: Proceedings of the 18th IEEE International Conference on Communication Technology (ICCT). Chongqing, China: IEEE, 2018. 911−915
[8]	Wang T, Wei L H, Ai J Q. Improved BP Neural Network for Intrusion Detection Based on AFSA. In: Proceedings of the 2015 International Symposium on Computers and Informatics (ISCI). Beijing, China: Atlantis Press, 2015. 373−380
[9]	Huang G, Song S J, Gupta J N D, Wu C. Semi-supervised and unsupervised extreme learning machines. IEEE Transactions on Cybernetics, 2014, 44(12): 2405-2417 doi: 10.1109/TCYB.2014.2307349
[10]	陆慧娟, 安春霖, 马小平, 郑恩辉, 杨小兵. 基于输出不一致测度的极限学习机集成的基因表达数据分类. 计算机学报, 2013, 36(2): 341-348 Lu Hui-Juan, An Chun-Lin, Ma Xiao-Ping, Zheng En-Hui, Yang Xiao-Bing. Disagreement measure based ensemble of extreme learning machine for gene expression data classification. Chinese Journal of Computers, 2013, 36(2): 341-348
[11]	陈晓云, 廖梦真. 基于稀疏和近邻保持的极限学习机降维. 自动化学报, 2019, 45(2): 325-333 Chen Xiao-Yun, Liao Meng-Zhen. Dimensionality reduction with extreme learning machine based on sparsity and neighborhood preserving. Acta Automatica Sinica, 2019, 45(2): 325-333
[12]	Yang Z X, Wang X B, Wong P K, Zhong J H. ELM based representational learning for fault diagnosis of wind turbine equipment. Proceedings of ELM-2015 Volume 2: Theory, Algorithms and Applications (II). Cham: Springer, 2016. 169−178
[13]	邹伟东, 夏元清. 基于压缩动量项的增量型ELM虚拟机能耗预测. 自动化学报, 2019, 45(7): 1290-1297 Zou Wei-Dong, Xia Yuan-Qing. Virtual machine power prediction using incremental extreme learning machine based on compression driving amount. Acta Automatica Sinica, 2019, 45(7): 1290-1297
[14]	Ku J H, Zheng B. Intrusion detection based on self-adaptive differential evolution extreme learning machine with gaussian kernel. In: Proceedings of the 8th International Symposium on Parallel Architecture, Algorithm and Programming. Haikou, China: Springer, 2017. 13−24
[15]	Deng W Y, Zheng Q H, Chen L. Regularized extreme learning machine. In: Proceedings of the 2009 IEEE Symposium on Computational Intelligence and Data Mining. Nashville, USA: IEEE, 2009. 389−395
[16]	Huang G B, Wang D H, Lan Y. Extreme learning machines: A survey. International Journal of Machine Learning and Cybernetics, 2011, 2(2): 107-122 doi: 10.1007/s13042-011-0019-y
[17]	Jiang X Y, Li S. BAS: Beetle antennae search algorithm for optimization problems [Online], available: https://arxiv.org/pdf/1710.10724.pdf, March 27, 2019
[18]	Jiang X Y. Li S. Beetle antennae search without parameter tuning (BAS-WPT) for multi-objective optimization. Filomat, 2020, 34(15): 5113−5119
[19]	刘影, 钱志鸿, 贾迪. 室内环境中基于天牛须寻优的普适定位方法. 电子与信息学报, 2019, 41(7): 1565-1571 doi: 10.11999/JEIT181021 Liu Ying, Qian Zhi-Hong, Jia Di. Universal localization algorithm based on beetle antennae search in indoor environment. Journal of Electronics & Information Technology, 2019, 41(7): 1565-1571 doi: 10.11999/JEIT181021
[20]	Wu Q, Ma Z P, Xu G, Li S, Chen D C. A novel neural network classifier using beetle antennae search algorithm for pattern classification. IEEE Access, 2019, 7: 64686-64696 doi: 10.1109/ACCESS.2019.2917526
[21]	Kaur G, Arora S. Chaotic whale optimization algorithm. Journal of Computational Design and Engineering, 2018, 5(3): 275-284 doi: 10.1016/j.jcde.2017.12.006
[22]	Ling Y, Zhou Y Q, Luo Q F. Lévy flight trajectory-based whale optimization algorithm for global optimization. IEEE Access, 2017, 5: 6168-6186 doi: 10.1109/ACCESS.2017.2695498
[23]	Sarangi A, Samal S, Sarangi S K. Analysis of Gaussian and Cauchy mutations in modified particle swarm optimization algorithm. In: Proceedings of the 5th International Conference on Advanced Computing and Communication Systems (ICACCS). Coimbatore, India: IEEE, 2019. 463−467
[24]	Rudolph G. Local convergence rates of simple evolutionary algorithms with Cauchy mutations. IEEE Transactions on Evolutionary Computation, 1997, 1(4): 249-258 doi: 10.1109/4235.687885
[25]	Gauthama Raman M R, Somu N, Kirthivasan K, Liscano R, Shankar Sriram V S. An efficient intrusion detection system based on hypergraph-Genetic algorithm for parameter optimization and feature selection in support vector machine. Knowledge-Based Systems, 2017, 134: 1-12 doi: 10.1016/j.knosys.2017.07.005
[26]	Mazini M, Shirazi B, Mahdavi I. Anomaly network-based intrusion detection system using a reliable hybrid artificial bee colony and AdaBoost algorithms. Journal of King Saud University-Computer and Information Sciences, 2019, 31(4): 541-553 doi: 10.1016/j.jksuci.2018.03.011
[27]	王东风, 孟丽. 粒子群优化算法的性能分析和参数选择. 自动化学报, 2016, 42(10): 1552-1561 doi: 10.16383/j.aas.2016.c150774 Wang Dong-Feng, Meng Li. Performance analysis and parameter selection of PSO algorithms. Acta Automatica Sinica, 2016, 42(10): 1552-1561 doi: 10.16383/j.aas.2016.c150774
[28]	UCI. UCI dataset [Online], available: https://archive.ics.uci.edu/ml/index.php, June 27, 2019
[29]	NSL-KDD dataset [Online], available: https://www.unb.ca/cic/datasets/nsl.html, June 27, 2019
[30]	Vinayakumar R, Alazab M, Soman K P, Poornachandran P, Al-Nemrat A, Venkatraman S. Deep learning approach for intelligent intrusion detection system. IEEE Access, 2019, 7: 41525-41550 doi: 10.1109/ACCESS.2019.2895334