基于模糊理论和BRBPNN的信息安全风险评估  

Information Security Risk Assessment Based on Fuzzy Theory and BRBPNN

作　　者：王鑫 唐作其[1] 许硕 WANG Xin;TANG Zuo-qi;XU Shuo(College of Computer Science&Technology,Guizhou University,Guiyang Guizhou 550025,China)

机构地区：[1]贵州大学计算机科学与技术学院

出　　处：《计算机仿真》2019年第11期184-189,共6页Computer Simulation

基　　金：贵州省科技计划课题(黔科合SY字[2011]3111);贵州大学青年教师科研基金项目(贵大青合字(2013)01号)

摘　　要：针对信息安全风险评估中存在大量不明确信息以及使用神经网络进行预测时会出现过拟合现象,提出了基于模糊理论和BRBPNN(贝叶斯正则化BP神经网络)相结合的信息安全风险评估方法。首先构建风险评估指标体系,建立风险评估模型,其次使用模糊理论处理原始数据,最后通过BR算法对BP神经网络进行训练。以某组织的信息系统为例,通过实验对比,表明在训练神经网络方面,BR算法比传统LM算法具有更好的健壮性和泛化能力。故上述方法为信息安全风险评估提供了新思路。Aimed at a large number of ambiguous information in the process of information security risk assessment and overfitting problem when using neural network to predict risk, an information security risk evaluation method based on fuzzy theory and Bayesian regularization BP neural network(BRBPNN) was proposed in this paper. Firstly, the risk assessment index system and the risk assessment model were established. Secondly, fuzzy theory was applied to process the original data. Finally, the BR algorithm was used to train the BP neural network. Taking an organization’s information system as an example, the experimental results showed that the neural network trained by BR algorithm has better robustness and generalization ability than the traditional LM algorithm. Therefore, this method provides novel ideas for information security risk assessment.

关 键 词：信息安全 风险评估 模糊理论 贝叶斯正则化 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]