作者：杨雅彬1, 刘晴1, 武志成2, 袁芬3
作者单位：1. 天津工业大学，天津 300387;
2. 杭州电子科技大学计算机学院，浙江 杭州 310018;
3. 浙江长征职业技术学院计算机与信息技术系，浙江 杭州 310023
关键词：机器学习;分布式网络;异常数据;智能检测;贝叶斯分类;遗传算法
摘要：
对于链路状态数据库的网络传输异常数据检测存在检测数据不完整、较为敏感、检测效率差的问题，提出基于机器学习的分布式网络传输异常数据智能检测方法，通过K最近邻分簇算法对分布式网络节点实施分簇，利用贝叶斯分类算法检测簇头是否出现异常；确定异常簇后，选取小波阈值降噪方法对异常簇内数据进行降噪处理，在此基础上，采用遗传算法检测降噪处理后异常簇内的异常数据，通过群体内最佳个体与最差个体的适应度函数值的差值同既定阈值的比较结果得到最终异常数据。经实验证明，所提方法检测异常数据的平均时间为8.48 s，检测结果与实际结果相似性较高，且检测性能较为稳定，说明该方法具有较高的异常数据检测性能。
Intelligent detection method for distributed network transmission abnormal data based on machine learning
YANG Yabin1, LIU Qing1, WU Zhicheng2, YUAN Fen3
1. Tiangong University, Tianjin 300387, China;
2. Computer & Software School, Hangzhou Dianzi University, Hangzhou 310018, China;
3. Department of Computer and Information Technology, Zhejiang Changzheng Vocational and Technical College, Hangzhou 310023, China
Abstract: At present, there are some problems in the detection of abnormal network transmission data in link state database, such as incomplete, sensitive and poor detection efficiency. Based on this, an intelligent detection method of abnormal data in distributed network transmission based on machine learning is proposed. The nodes in the distributed network are clustered by K-nearest neighbor clustering algorithm, and whether the cluster head is abnormal is detected by Bayesian classification algorithm. After determining the abnormal cluster, the wavelet threshold denoising method is selected to denoise the abnormal data in the cluster. On this basis, genetic algorithm is used to detect the abnormal data after noise reduction, the abnormal data in the abnormal cluster is measured, and the final abnormal data is obtained by comparing the difference between the fitness function value of the best individual and the worst individual in the group with the given threshold value. The experimental results show that the average time of detecting abnormal data is about 8.48 s, the detection results are similar to the actual results, and the detection performance is relatively stable, which shows that the method has high performance of abnormal data detection.
Keywords: machine learning;distributed network;abnormal data;intelligent detection;Bayesian classification;genetic algorithm
2021, 47(3):104-109 收稿日期: 2020-02-19;收到修改稿日期: 2020-03-18
基金项目:
作者简介: 杨雅彬(1991-),男,天津市人,助理实验师,硕士,专业方向为信息化技术
参考文献
[1] 许春杰, 吴蒙, 杨立君. 一种基于分层聚合的分布式异常数据检测方案[J]. 计算机工程, 2020, 46(4): 213-219
[2] 黄军伟, 唐娴. 网络信息传输异常数据检测仿真研究[J]. 计算机仿真, 2018, 35(10): 398-401
[3] 许刚, 王展, 臧大伟, 等. 基于链路状态数据库的数据中心网络异常检测算法[J]. 计算机研究与发展, 2018, 55(4): 815-830
[4] 靳紫辉, 夏钰红. 云计算下分布式大数据智能融合算法仿真[J]. 计算机仿真, 2018, 35(10): 295-298
[5] YIN L, YE B, ZHANG Z, et al. A novel feature extraction method of eddy current testing for defect detection based on machine learning[J]. NDT & E international, 2019, 107(10): 102108.1-102108.7
[6] UCAR F, CORDOVA J, ALCIN O F, et al. Bundle extreme learning machine for power quality analysis in transmission networks[J]. Energies, 2019, 12(8): 1449-1449
[7] WU S, LI D Z, WANG Z Y, et al. Novel distributed UEP rateless coding scheme for data transmission in deep space networks[J]. Science China(Information Sciences), 2018, 61(4): 87-96
[8] JIN Y, XIA K J. Intelligent location and recognition mechanism of abnormal point of medical image based on reliable transmission in medial CT scanner local big data networks[J]. Journal of Medical Imaging and Health Informatics, 2018, 8(3): 609-617
[9] 赖清, 曾红武. 网络数据异常信息流量传输安全性检测仿真[J]. 计算机仿真, 2018: 293-296
[10] 丘洪伟, 李小映. 关于交互式网络中异常数据准确检测仿真[J]. 计算机仿真, 2018, 35(5): 375-378
[11] 唐家琪, 吴璟莉. 基于PPI网络与机器学习的蛋白质功能预测方法[J]. 计算机应用, 2018, 38(3): 722-727
[12] 李小玲. 关于网络数据库传输中异常数据检测仿真研究[J]. 计算机仿真, 2018, 35(1): 420-423
[13] 陈铁明, 金成强, 吕明琪, 等. 基于样本增强的网络恶意流量智能检测方法[J]. 通信学报, 2020, 41(6): 128-138
