基于改进鲸鱼算法优化SVM的软件缺陷检测方法

doi:10.3969/j.issn.1671-1122.2024.08.002

摘要/Abstract

摘要：

为解决传统支持向量机在软件缺陷检测中存在分类精度低、参数选择困难等问题，文章提出一种基于改进鲸鱼算法优化SVM的软件缺陷检测方法LFWOA-SVM。首先针对鲸鱼算法在求解过程中存在收敛速度慢、寻优效率低和局部最优解问题，基于Levy飞行策略优化鲸鱼觅食阶段，最大限度地实现搜索代理多样化，并利用混合变异扰动算子提高WOA的全局寻优能力；然后采用改进的鲸鱼算法LFWOA对SVM的惩罚因子和核函数参数进行优化，在获得最优参数的同时可有效检测软件缺陷。仿真实验表明，在6个基准测试函数中，LFWOA展现出更高的寻优速度和全局搜索能力；在8个公开软件缺陷数据集上进行测试显示，LFWOA-SVM方法能够有效提高分类性能和预测精度。

关键词: 软件缺陷检测, Levy飞行, 鲸鱼优化算法, 变异扰动, 支持向量机

Abstract:

To enhance the performance of software defect detection, a refined model called LFWOA-SVM has been proposed, utilizing an improved Whale algorithm to optimize traditional SVM. This approach aims at inherent issues of SVM, such as low classification accuracy and complex parameter tuning. First, in view of the problems of slow convergence speed, low optimization efficiency and local optimal solution in the whale algorithm during the solution process, the whale foraging stage was optimized based on the levy flight strategy to maximize the diversification of search agents, and a hybrid mutation perturbation was proposed operators were used to improve WOA’s global optimization capabilities. Secondly, the improved whale algorithm LFWOA was used to optimize the penalty factor and kernel function parameters of SVM, which can be effectively used in software defect detection while obtaining the optimal parameters. Finally, data simulation experiments show that among 6 benchmark test functions, LFWOA exhibits higher optimization speed and global search capabilities; tests on 8 public software defect data sets show that LFWOA-SVM method can effectively improve identification performance and prediction accuracy.

Key words: software defect detection, Levy flight, whale optimization algorithm, mutation perturbation, SVM

中图分类号:

TP309

杜晔, 田晓清, 李昂, 黎妹红. 基于改进鲸鱼算法优化SVM的软件缺陷检测方法[J]. 信息网络安全, 2024, 24(8): 1152-1162.

DU Ye, TIAN Xiaoqing, LI Ang, LI Meihong. Software Defect Detection Method Based on Improved Whale Algorithm to Optimize SVM[J]. Netinfo Security, 2024, 24(8): 1152-1162.

图/表 11

图1

图2

图3

图4

图5

表1

表2

表3

图6

表4

表5

参考文献 27

[1]	LI Minglei, HUANG Hui, LU Yuliang. Test Case Generation Technology Based on Symbolic Divide and Conquer Zones for Vulnerability Mining[J]. Netinfo Security, 2020, 20 (5): 39-46.
	李明磊, 黄晖, 陆余良. 面向漏洞挖掘的基于符号分治区的测试用例生成技术[J]. 信息网络安全, 2020, 20(5): 39-46.
[2]	ALSAEEDI A, KHAN M Z. Software Defect Prediction Using Supervised Machine Learning and Ensemble Techniques: A Comparative Study[J]. Journal of Software Engineering and Applications, 2019, 12(5): 85-100.
[3]	LI Sicong, WANG Jian, SONG Yafei, et al. Malicious Code Classification Method Based on BiTCN-DLP[J]. Netinfo Security, 2023, 23(11): 104-117.
	李思聪, 王坚, 宋亚飞, 等. 基于BiTCN-DLP的恶意代码分类方法[J]. 信息网络安全, 2023, 23(11): 104-117.
[4]	GRAY D, BOWES D, DAVEY N, et al. Using the Support Vector Machine as a Classification Method for Software Defect Prediction with Static Code Metrics[C]// Springer. International Conference on Engineering Applications of Neural Networks. Heidelberg: Springer, 2009: 223-234.
[5]	YANG Zhanyu, LU Lu, ZOU Quanyi. Ensemble Kernel-Mapping-Based Ranking Support Vector Machine for Software Defect Prediction[J]. IEEE Transactions on Reliability, 2024, 73(1): 664-679.
[6]	RAJABI A, BOBBA R B. False Data Detection in Distributed Oscillation Mode Estimation Using Hierarchical K-Means[C]// IEEE. 2019 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm). New York: IEEE, 2019: 1-7.
[7]	CHATTERJEE S, MAJI B. A Bayesian Belief Network Based Model for Predicting Software Faults in Early Phase of Software Development Process[J]. Applied Intelligence, 48(6): 1-15.
[8]	MANJULA C, FLORENCE L. Deep Neural Network Based Hybrid Approach for Software Defect Prediction Using Software Metrics[EB/OL]. [2024-03-10]. https://link.springer.com/article/10.1007/s10586-018-1696-z.
[9]	BORANDAG E, OZCIFT A, KILINC D, et al. Majority Vote Feature Selection Algorithm in Software Fault Prediction[J]. Computer Science and Information Systems, 2019, 16(2): 515-539. doi: 10.2298/CSIS180312039B
[10]	GRAY D, BOWES D, DAVEY N, et al. Using the Support Vector Machine as a Classification Method for Software Defect Prediction with Static Code Metrics[C]// Springer. International Conference on Engineering Applications of Neural Networks. Heidelberg: Springer, 2009, 223-234.
[11]	REN Shengbing, LIAO Xiangdang. Software Defect Prediction Based on Cost-Sensitive Support Vector Machine[J]. Computer Engineering & Science, 2018, 40(10): 1787-1795.
	任胜兵, 廖湘荡. 基于代价敏感支持向量机的软件缺陷预测研究[J]. 计算机工程与科学, 2018, 40(10): 1787-1795.
[12]	WANG Ruoping, CHEN Yan, WANG Dong, et al. Vehicle Interior Abnormal Noise Identification Method Based on Multi-Feature Extraction and Grey Wolf Algorithm Optimization SVM[J]. Computer Applications and Software, 2024, 41(3): 41-48.
	王若平, 陈严, 王东, 等. 基于多特征提取与灰狼算法优化SVM的车内异响识别方法[J]. 计算机应用与软件, 2024, 41(3): 41-48.
[13]	AZZEH M, NASSIF A B, BANITAAN S. On Selection of Optimal Kernel Function for Software Defect Prediction[C]// IEEE. 2022 21st IEEE International Conference on Machine Learning and Applications(ICMLA). New York: IEEE, 2022: 1761-1768.
[14]	ZHANG Xinying, LI Bin, WU Yuanyuan. Malware Detection Based on Synchronization Optimizing Feature Selection and Support Vector Machine[J]. Computer Engineering and Design, 2024, 45(2): 467-476.
	张新英, 李彬, 吴媛媛. 针对恶意软件检测的特征选择与SVM协同优化[J]. 计算机工程与设计, 2024, 45(2): 467-476.
[15]	FENG Zhang, PEI Dong, WANG Wei. Face Recognition by Support Vector Machine Optimized by an Improved Grey Wolf Algorithm[J]. Computer Engineering & Science, 2019, 41(6): 1057-1063.
	冯璋, 裴东, 王维. 基于改进灰狼算法优化支持向量机的人脸识别[J]. 计算机工程与科学, 2019, 41(6): 1057-1063.
[16]	ZHANG Ran, LIU Min, PAN Zhihan, et al. Network Security Situation Assessment Based on Improved WOA-SVM[J]. IEEE Access, 2022, 10: 96273-96283.
[17]	ACHARYA S, SIVARAJAN G, KUMAR D V, et al. Modeling Combined Economic Emission Dispatch for Renewable Energy System via Levy-Based Glowworm Swarm Optimization[J]. Kybernetes, 2023, 52(9): 3315-3337.
[18]	MIRJALILI S, LEWIS A. The Whale Optimization Algorithm[J]. Advances in Engineering Software, 2016, 95: 51-67.
[19]	WEI Zihui, LI Xiaoyang, WANG Le, et al. Improved TDOA Three-Dimensional Positioning Algorithm Based on Adaptive Levy Flight[J]. Journal of Hebei University(Natural Science Edition), 2023, 43(2): 207-215. doi: 10.3969/j.issn.1000-1565.2023.02.013
	韦子辉, 李小阳, 王勒, 等. 基于自适应Levy飞行改进的TDOA三维定位算法[J]. 河北大学学报(自然科学版), 2023, 43(2): 207-215. doi: 10.3969/j.issn.1000-1565.2023.02.013
[20]	MIRJALILI S, MIRJALILI S M, LEWIS A. Grey Wolf Optimizer[J]. Advances in Engineering Software, 2014, 69(3): 46-61.
[21]	OZCAN E, MOHAN C K. Particle Swarm Optimization: Surfing the Waves[C]// IEEE. The 1999 Congress on Evolutionary Computation-CEC99. New York: IEEE, 1999: 1939-1944.
[22]	HE Qing, WEI Kangyuan, XU Qinshuai. Improved Whale Optimization Algorithm Based on Mixed Strategy[J]. Computer Application Research, 2019, 36(12): 3647-3651, 3665.
	何庆, 魏康园, 徐钦帅. 基于混合策略改进的鲸鱼优化算法[J]. 计算机应用研究, 2019, 36(12): 3647-3651,3665.
[23]	TANG Yu, DAI Qi, YANG Zhiwei, et al. Research on Software Defect Prediction Algorithm Based on Optimized Random Forest[J]. Computer Engineering and Science, 2023, 45(5): 830-839.
	唐宇, 代琪, 杨志伟, 等. 基于优化随机森林的软件缺陷预测算法研究[J]. 计算机工程与科学, 2023, 45(5): 830-839.
[24]	MANCHALA P, BISI M. Diversity Based Imbalance Learning Approach for Software Fault Prediction Using Machine Learning Models[EB/OL]. [2024-03-20]. https://www.sciencedirect.com/science/article/abs/pii/S1568494622003660?via%3Dihub.
[25]	SHIRABAD J S, MENZIES T J. The PROMISE Repository of Software Engineering Databases[EB/OL]. [2024-03-20]. https://www.mendeley.com/catalogue/016e9979-cca5-3f49-a0ac-abc6a86570ef/.
[26]	SHEPPERD M, SONG Qinbao, SUN Zhongbin, et al. Data Quality: Some Comments on the NASA Software Defect Datasets[J]. IEEE Transactions on Software Engineering, 2013, 39(9): 1208-1215.
[27]	XIONG Zhongmin, GUO Huaiyu, WU Yuexin. Review of Research on Missing Data Processing Methods[J]. Computer Engineering and Applications, 2021, 57(14): 27-38. doi: 10.3778/j.issn.1002-8331.2101-0187
	熊中敏, 郭怀宇, 吴月欣. 缺失数据处理方法研究综述[J]. 计算机工程与应用, 2021, 57(14): 27-38. doi: 10.3778/j.issn.1002-8331.2101-0187

类型	函数	维度	搜索范围
单峰	${{F}_{1}}(x)=\sum\limits_{i=1}^{n}{x_{i}^{2}}$	30	[-100,100]
	${{F}_{2}}(x)=\sum\limits_{i=1}^{n}{{{\left( \sum\limits_{j=1}^{n}{{{x}_{j}}} \right)}^{2}}}$	30	[-30,30]
	${{F}_{3}}(x)=\max \left\{ \left\| {{x}_{i}} \right\| \right\}$	30	[-100,100]
多峰	$\begin{align} & {{F}_{4}}(x)=-20\exp \left( -0.2\sqrt{\frac{1}{n}\sum\limits_{i=1}^{n}{\left\| x_{i}^{2} \right\|}} \right)+20+e \\ & \ \ -\exp \left( \frac{1}{n}\sum\limits_{i=1}^{n}{\cos (2\text{ }\!\!\pi\!\!\text{ }{{x}_{i}})} \right) \end{align}$	30	[-32,32]
	${{F}_{5}}(x)=\frac{1}{4000}\sum\limits_{i=1}^{n}{x_{i}^{2}}-\prod\limits_{i=1}^{n}{\cos \left( \frac{{{x}_{i}}}{\sqrt{i}} \right)+1}$	30	[-600,600]
	${{F}_{6}}(x)=\frac{1}{4000}\sum\limits_{i=1}^{n}{\left[ x_{i}^{2}-10\cos (2\text{ }\!\!\pi\!\!\text{ }{{x}_{i}})+10 \right]}$	30	[-5.12,5.12]

数据集	特征数/个	数据量/个	类别
MC2	39	125	2
PC1	21	1108	2
KC3	39	458	2
MW1	39	403	2
PC3	37	1053	2
tomcat	20	858	2
jedit-4.2	20	367	2
camel-1.6	20	965	2

基准函数	算法	均值	标准差	最优解
F₁	LFWOA	0.0000E+00	0.0000E+00	0.0000E+00
	GWO	1.1788E-27	1.4877E-27	2.8560E-29
	WOA	2.6161E-28	6.4843E-28	1.0315E-32
	PSO	8.4296E-01	3.6406E-01	2.2663E-01
	MS-WOA	0.0000E+00	0.0000E+00	0.0000E+00
F₂	LFWOA	0.0000E+00	0.0000E+00	0.0000E+00
	GWO	1.8138E-05	3.5241E-05	9.9501E-09
	WOA	6.6487E+02	6.8947E+02	2.3600E+01
	PSO	7.2326E+02	8.1981E+02	6.2506E+01
	MS-WOA	7.3821E-77	2.9200E-76	6.4254E-88
F₃	LFWOA	0.0000E+00	0.0000E+00	0.0000E+00
	GWO	6.7930E-07	6.7869E-07	4.9692E-08
	WOA	5.7282E-02	9.7826E-02	4.0602E-04
	PSO	3.7070E+00	1.3278E+00	1.5407E+00
	MS-WOA	8.7712E-48	2.8337E-47	1.2352E-52
F₄	LFWOA	0.0000E+00	0.0000E+00	0.0000E+00
	GWO	3.0885E+00	3.9162E+00	5.6843E-14
	WOA	6.2138E+01	4.2734E+01	1.9973E+00
	PSO	5.2677E+01	1.2228E+01	3.4609E+01
	MS-WOA	0.0000E+00	0.0000E+00	0.0000E+00
F₅	LFWOA	0.0000E+00	0.0000E+00	8.8818E-16
	GWO	5.6843E-14	5.6843E-14	7.5495E-14
	WOA	1.9973E+00	1.9973E+00	2.2204E-14
	PSO	3.4609E+01	3.4609E+01	3.3968E+00
	MS-WOA	0.0000E+00	0.0000E+00	8.8818E-16
F₆	LFWOA	0.0000E+00	0.0000E+00	0.0000E+00
	GWO	5.8887E-03	9.2274E-03	0.0000E+00
	WOA	2.4726E-03	6.7279E-03	0.0000E+00
	PSO	2.0872E+02	1.7547E+01	1.7169E+02
	MS-WOA	5.5940E-13	3.0640E-12	0.0000E+00

数据集	SVM	WOA-SVM	PSO-SVM	LFWOA-SVM
MC2	0.538	0.600	0.567	0.638
PC1	0.581	0.664	0.663	0.671
MW1	0.500	0.643	0.639	0.643
KC3	0.500	0.538	0.542	0.551
PC3	0.757	0.762	0.675	0.764
tomcat	0.793	0.803	0.732	0.794
jedit-4.2	0.706	0.801	0.726	0.804
camel-1.6	0.611	0.631	0.597	0.673
均值	0.623	0.680	0.642	0.693

数据集	SVM	PSO-SVM	WOA-SVM	LFWOA-SVM
MC2	0.077	0.364	0.311	0.386
PC1	0.316	0.384	0.480	0.499
MW1	0.039	0.381	0.312	0.468
KC3	0.001	0.258	0.220	0.143
PC3	0.003	0.202	0.205	0.324
tomcat	0.002	0.147	0.116	0.152
jedit-4.2	0.220	0.401	0.411	0.415
camel-1.6	0.042	0.149	0.175	0.181
均值	0.086	0.286	0.279	0.321