The Information Security Risk Assessment Model Based On AFSA_SVM

Network security is becoming a hot issue that is studied and paid attention to. People estimate the network security situation focus on how to use expertise before, but now, how to assess the network security situation correctly and accurately at the real time becomes an important research directive of network security decision system. Based on theories and technologies of network security situation assessment, this thesis presents an automated assessment method, an assessment model which is based on AFSA-SVM. The following studies are done in this thesis: 1. The paper statistics all security issues of network security situation nationwide,analyzes and sorts out new sample sets of influential factors. It ensures thenetwork security influential factors aims at government Versa Net. 2. The paper selects classified model of SVM, and experimentally compares thealgorithm performance of SVM’s commonly used kernel function. It finds optimalkernel function of RBF for SVM classifier in this area. 3. The paper applies the method of AFSA parameter optimization to the process ofSVM algorithm parameter optimization, finds the optimal parameter c and gthrough the process of AFSA’s global optimization, and improves the velocity aprecision of parameter optimization at the same time. 4. The paper proves AFSA optimizing is more advanced than Cross-validation optimizing and PSO(particle swarm algorithm) optimizing in velocity andprecision through experimental comparison. 5. The paper engineering realizes assessment model, and achieves network securitysituation assessment model of network security decision system. It visualprograms the network security decision model and assessment model, which playsa positive role in decision process.This thesis establishes a new sample aims to government network security situation assessment. Meanwhile, it applies a new algorithm to improve the velocity and precision of assessment. It realizes engineering application of automated assessment model and visualization of decision system. Experiments show this model can assess current network security situation quickly, accurately and timely.