Quantitative Research On Bayesian Neural Network In Information Security Risk Assessment

Nowadays frequent information security accidents not only have caused great harm to information system but also have brought many thorny problems.Therefore guaranteeing the security of information system remains an urgent problem.Quantitative risk assessment on information system,as one of important tools to achieve information security,is of great theoretical and practical significance,which has attracted many scholars' attention.However,the traditional quantitative risk assessment is time-consuming and laborious,and the evaluation results are influenced by some subjective factors.Aimed at those issues,this study mainly involves the following aspects:Firstly,this study puts forward a method of quantified risk based on fuzzy theory and Bayesian regularized BP neural network.This method can decrease the fuzziness of language while artificial neural networks technology speed up the speed of quantifying the risks.Using fuzzy theory to process the raw data,and using the processed data as the input of the neural network,can effectively reduce the ambiguity in the language description.At the same time,the Bayesian regularization algorithm is used to train the neural network because it is easy to fall into the local optimal problem.Then the effectiveness of risk quantification method is verified by experimental simulation;Secondly,safety protection measures and safety precautions are proposed as new indicator in this paper.By adding new indicators,a more comprehensive information security risk assessment indicator system suitable for information systems can be constructed.Through experimental verification,adding new indicators can more fully describe the current state of system risk,and has strong practicability.The findings provide strong support for enterprises and organizations to carry out further risk improvement programs.