中国邮电高校学报(英文版) ›› 2017, Vol. 24 ›› Issue (3): 33-43.doi: 10.1016/S1005-8885(17)60209-7

• Networks • 上一篇    下一篇

Network security situation automatic prediction model based on accumulative CMA-ES optimization

王健,李可   

  1. 哈尔滨理工大学
  • 收稿日期:2016-12-30 修回日期:2017-04-07 出版日期:2017-06-30 发布日期:2017-06-30
  • 通讯作者: 王健 E-mail:wangjianlydia@163.com
  • 基金资助:
    国家自然科学基金重点项目;国家自然科学基金重点项目;高等学校博士学科点专项科研基金

Network security situation automatic prediction model based on accumulative CMA-ES optimization

2   

  • Received:2016-12-30 Revised:2017-04-07 Online:2017-06-30 Published:2017-06-30
  • Supported by:
    National Science Foundation of China;National Science Foundation of China

摘要: To improve the accuracy of the network security situation, a security situation automatic prediction model based on accumulative data preprocess and support vector machine (SVM) optimized by covariance matrix adaptive evolutionary strategy (CMA-ES) is proposed. The proposed model adopts SVM which has strong nonlinear ability. Also, the hyper parameters for SVM are optimized through the CMA-ES which owns good performance in finding optimization automatically. Considering the irregularity of network security situation values, we accumulate the original sequence, so that the internal rules of discrete data can be revealed and it is easy to model. Simulation experiments show that the proposed model has faster convergence-speed and higher prediction accuracy than other extant prediction models.

关键词: security situation, automatic prediction, covariance matrix adaptive evolution strategy, support vector machine

Abstract: To improve the accuracy of the network security situation, a security situation automatic prediction model based on accumulative data preprocess and support vector machine (SVM) optimized by covariance matrix adaptive evolutionary strategy (CMA-ES) is proposed. The proposed model adopts SVM which has strong nonlinear ability. Also, the hyper parameters for SVM are optimized through the CMA-ES which owns good performance in finding optimization automatically. Considering the irregularity of network security situation values, we accumulate the original sequence, so that the internal rules of discrete data can be revealed and it is easy to model. Simulation experiments show that the proposed model has faster convergence-speed and higher prediction accuracy than other extant prediction models.

Key words: security situation, automatic prediction, covariance matrix adaptive evolution strategy, support vector machine