不平衡数据下基于SVM增量学习的指挥信息系统状态监控方法

doi:10.12305/j.issn.1001-506X.2024.03.25

摘要/Abstract

摘要：

针对指挥信息系统历史状态样本有限的特点, 基于支持向量机(support vector machines, SVM)设计了一种面向不平衡数据的SVM增量学习方法。针对系统正常/异常状态样本不平衡的情况, 首先利用支持向量生成一部分新样本, 然后通过分带的思想逐带产生分布更加均匀的新样本以调节原样本集的不平衡比。针对系统监控实时性要求高且在运行过程中会有新样本不断加入的特点, 采用增量学习的方式对分类模型进行持续更新, 在放松KKT(Karush-Kuhn-Tucker)更新触发条件的基础上, 通过定义样本重要度并引入保留率和遗忘率的方式减少了增量学习过程中所需训练的样本数量。为了验证算法的有效性和优越性, 实验部分在真实系统中获得的数据集以及UCI数据集中3类6组不平衡数据集中与现有的算法进行了对比。结果表明, 所提算法能够有效实现对不平衡数据的增量学习, 从而满足指挥信息系统状态监控的需求。

关键词: 指挥信息系统, 系统监控, 支持向量机, 不平衡数据, 增量学习

Abstract:

To the characteristic of limited historical sample of command, control, communication, and computer, intelligence, surveillance and reconnaissance (C4ISR), an incremental learning method based on support vector machines (SVM) is designed for imbalanced data. To the imbalance of normal/abnormal state samples of the system, first use the support vector to generate a part of new samples, and then use the idea of banding to generate new samples with a more uniform distribution to adjust the imbalance ratio of the original sample set. In view of the high requirements for real-time monitoring of the system and the continuous addition of new samples during operation, the classification model is continuously updated by incremental learning. On the basis of relaxing the KKT (Karush-Kuhn-Tucker) update triggering conditions, by defining the sample importance and the introduction of retention rate/forgetting rate to reduce the number of training samples required in the incremental learning process. In order to verify the effectiveness and superiority of the algorithm, the experimental part compared the existing algorithms in the real system data set and the UCI data set with 3 types and 6 groups of imbalanced data sets. The results show that the proposed algorithm can effectively realize the incremental learning of imbalanced data, so as to meet the requirements of the C4ISR state monitoring.

Key words: command control communication and computer intelligence surveillance and reconnaissance (C4ISR), system monitoring, support vector machine (SVM), imbalanced data, incremental learning

中图分类号:

TP315

焦志强, 易侃, 张杰勇, 姚佩阳. 不平衡数据下基于SVM增量学习的指挥信息系统状态监控方法[J]. 系统工程与电子技术, 2024, 46(3): 992-1003.

Zhiqiang JIAO, Kan YI, Jieyong ZHANG, Peiyang YAO. C4ISR state monitoring method based on SVM incremental learning of imbalanced data[J]. Systems Engineering and Electronics, 2024, 46(3): 992-1003.

图/表 17

图1

图2

图3

图4

图5

图6

图7

图8

图9

表1

表2

表3

图10

表4

表5

图11

图12

参考文献 30

1	张兆晨, 周光霞. 指挥信息系统技术发展趋势分析[C]//第六届中国指挥控制大会, 2018(1): 52-56.
	ZHANG Z C, ZHOU G X. Analysis of technology development trend of command information[C]//Proc. of the 6th China Command and Control Conference, 2018(1): 52-56.
2	ZAVALA E. Towards adaptive monitoring services for self-adaptive software systems[C]//Proc. of the International Conference on Service Oriented Computing, 2017: 357-362.
3	DEVRIES B, CHENG B H. Run-time monitoring of self-adaptive systems to detect N-way feature interactions and their causes[C]//Proc. of the Software Engineering for Adaptive And Self Managing Systems, 2018: 94-100.
4	CHENG W, LI Q S, WANG L, et al. Handling uncertainty online for self-adaptive systems[C]//Proc. of the 5th International Conference on Soft Computing & Machine Intelligence, 2018: 135-139.
5	ZHANG H, LI Q S, WANG L, et al. Self-adaptive software changes analysis method based on "detection-recognition" mechanism(S)[C]//Proc. of the 31st International Conference on Software Engineering and Knowledge Engineering, 2019: 287-378.
6	吴俊锋, 毛杰, 丁健. 指挥信息系统一站式综合监控系统设计与实现[J]. 指挥信息系统与技术, 2013, 4 (5): 61- 66.
	WU J F , MAO J , DING J . Design and implementation of one-stop unified monitoring system for command information system[J]. Command Information System and Technology, 2013, 4 (5): 61- 66.
7	张超梅, 金晓雪, 戴仔强. 复杂信息系统的故障诊断专家系统设计与实现[J]. 指挥信息系统与技术, 2013, 4 (8): 27- 32.
	ZHANG C M , JIN X X , DAI Z Q . Design and realization of fault diagnosis expert system in complex information system[J]. Command Information System and Technology, 2013, 4 (8): 27- 32.
8	梁建兴, 贾奖, 唐昌建. 基于WinPcap的指控网流量监控方法研究[C]//第二届中国指挥控制大会, 2014: 600-602.
	LIANG J X, JIA J, TAN C J. Research on traffic monitoring method based on WinPcap[C]//Proc. of the 2nd China Command and Control Conference, 2014: 600-602.
9	DEAN D J, NGUYEN H, GU X. UBL: unsupervised behavior learning for predicting performance anomalies in virtualized cloud systems[C]//Proc. of the International Conference on Autonomic Computing, 2012.
10	ZHANG Y C, HONG B, ZHANG M, et al. ECAD: cloud anomalies detection from an evolutionary view[C]//Proc. of the International Conference on Cloud Computing & Big Data, 2013.
11	ACOSTA J A , GONZALEZ J P , FIGUEROA V M , et al. ICA-multiclass SVM as a monitoring system of complex processes[J]. Research on Computing Science, 2014, 82 (1): 41- 52. doi: 10.13053/rcs-82-1-4
12	SU Y, ZHANG W Z, TAO W W, et al. A network illegal access detection method based on PSO-SVM algorithm in power monitoring system[C]//Proc. of the International Conference on Cloud Computing, 2018: 450-459.
13	SUNDARAVADIVEL P , KESAVAN K , KESAVAN L , et al. Smart-log: a deep-learning based automated nutrition monitoring system in the IoT[J]. IEEE Trans.on Consumer Electronics, 2018, 64 (3): 390- 398. doi: 10.1109/TCE.2018.2867802
14	JI C Z , LU S Q . Exploration of marine ship anomaly real-time monitoring system based on deep learning[J]. Journal of Intelligent and Fuzzy Systems, 2020, 38 (2): 1235- 1240. doi: 10.3233/JIFS-179485
15	唐庭龙. 支持向量机增量学习研究[D]. 杭州: 浙江工业大学, 2018.
	TANG T L. Reserarch on support vector machine incremental learning[D]. Hangzhou: Zhejiang University of Technology, 2018.
16	YU H L , NI J , ZHAO J . ACOS ampling: an ant colony optimization-based under sampling method for classifying imbalanced DNA microarray data[J]. Neurocomputing, 2013, 101 (2): 309- 318.
17	ARYA I , HAMED M S , NUNO V . Cost sensitive support vector machines[J]. Neurocomputing, 2019, 343 (5): 50- 64.
18	赵小强, 张露. 基于SVM的高维不平衡数据集分类算法[J]. 南京大学学报(自然科学), 2018, 54 (2): 238- 247.
	ZHAO X Q , ZHANG L . Classification algorithm of high-dimensional and imbalanced data based on support vector machine[J]. Journal of NanJing University (Natural Science), 2018, 54 (2): 238- 247.
19	黄海松, 魏建安, 康佩栋. 基于不平衡数据样本特性的新型过采样SVM分类算法[J]. 控制与决策, 2018, 33 (9): 1549- 1558.
	HUANG H S , WEI J A , KANG P D . New over-sampling SVM classification algorithm based on unbalanced data sample characteristics[J]. Control and Decision, 2018, 33 (9): 1549- 1558.
20	CHENG W Y , JUANG C F . An incremental support vector machine-trained TS-type fuzzy system for online classification problems[J]. Fuzzy Sets and Systems, 2011, 163 (1): 24- 44. doi: 10.1016/j.fss.2010.08.006
21	YANG Y L , CHE J X , LI Y Y , et al. An incremental electric load forecasting model based on support vector regression[J]. Energy, 2016, 113 (10): 796- 808.
22	YI Y , WU J S , XU W , et al. Incremental SVM based on reserved set for network intrusion detection[J]. Expert Systems With Applications, 2011, 38 (6): 7698- 7707. doi: 10.1016/j.eswa.2010.12.141
23	CHITRAKAR R , HUANG C . Selection of candidate support vectors in incremental SVM for network intrusion detection[J]. Computers & Security, 2014, 45 (9): 231- 241.
24	LI C H , LIU K W , WANG H X , et al. The incremental learning algorithm with support vector machine based on hyperplane-distance[J]. Applied Intelligence, 2011, 34 (1): 19- 27.
25	KUBAT M, HOLTE R C, MATWIN S, et al. Learning when negative examples abound[C]//Proc. of the European Confe-rence on Machine Learning, 1997: 146-153.
26	DAVIS J, GOADRICH M. The relationship between precision-recall and ROC curves[C]//Proc. of the International Confe-rence on Machine Learning, 2006: 233-240.
27	LIN K C, CHIEN H Y. CSO-based feature selection and parameter optimization for support vector machine[C]//Proc. of the Joint Conferences on Pervasive Computing, 2009: 783-788.
28	YANG Z M, YANG Y, GANG W. Classification for imba-lanced dataset based on biased empirical feature mapping[C]//Proc. of the IEEE International Instrumentation and Measurement Technology Conference, 2012: 1645-1649.
29	KHAN M, ARIF R B, SIDDIQUE M, et al. Study and observation of the variation of accuracies of KNN, SVM, LMNN, ENN algorithms on eleven different datasets from UCI machine learning repository[C]//Proc. of the 4th International Confe-rence on Electrical Engineering and Information & Communication Technology, 2018: 124-129.
30	SUN C, YUE S H, LI Q. Clustering characteristics of UCI dataset[C]//Proc. of the 39th Chinese Control Conference, 2020: 6301-6306.

算法	0	1	2	3	4	5	6	7	8	9	10
SVM	200.0	220.0	240.0	260.0	280.0	300.0	320.0	340.0	360.0	380.0	400.0
SVM+INV	200.0	206.3	226.1	245.9	265.7	285.6	305.5	325.3	345.2	365.2	385.0
SVM+UB	300.0	323.9	347.9	371.7	395.8	419.9	443.9	468.0	492.0	516.0	540.0
本文算法	300.0	289.7	309.2	324.5	339.8	354.8	370.1	385.2	398.6	411.9	424.8

算法	0	1	2	3	4	5	6	7	8	9	10
SVM	0.35	2.99	2.97	2.87	2.90	3.00	3.22	3.37	3.47	3.52	3.54
SVM+INV	0.35	4.34	4.36	4.48	4.51	4.75	4.92	5.10	5.24	5.36	5.54
SVM+UB	8.81	12.90	13.72	14.50	15.69	16.38	17.87	18.77	20.10	21.70	22.82
本文算法	8.88	13.03	13.47	14.16	14.67	15.14	15.29	16.38	16.48	16.97	17.37

数据集	特征数目	样本数量	不平衡比
yeast1	8	1 484	2.06
yeast3	8	1 484	8.1
ecoli1	7	336	3.36
ecoli4	7	336	15.8
poker-8-9vs6	10	1 485	58.4
poker-8-9vs5	10	2 075	82

算法	yeast1	yeast3	ecoli1	ecoli4	poker-8-9vs6	poker-8-9vs5
Simple-ISVM	332.0	175.4	59.1	51.3	99.5	124.8
KKT-ISVM	486.6	207.0	71.7	45.9	56.3	62.0
CRS-ISVM	486.8	206.9	73.8	46.8	61.3	66.4
本文算法	659.5	657.9	119.0	132.5	568.4	797.5

算法	yeast1	yeast3	ecoli1	ecoli4	poker-8-9vs6	poker-8-9vs5
Simple-ISVM	4.44	2.69	1.77	1.78	2.07	2.21
KKT-ISVM	8.36	3.86	1.95	1.88	2.44	2.66
CRS-ISVM	11.16	6.18	2.52	2.48	4.84	5.89
本文算法	36.24	50.38	4.01	5.13	42.42	77.33