Research On Defect Prediction Model Of Internet Of Things System Based On Machine Learning

With the rapid development of the Internet of things and the rise of intelligent terminals,a large number of intelligent terminals are connected to the Internet.However,due to the lack of security awareness of the Internet of things manufacturers and less investment in security research,the development and implementation speed of security standards is far lower than the development speed of the Internet of things industry,which leads to a great potential risk in some Internet of things systems.Some illegal personnel use the defects or loopholes of the Internet of things terminals to attack,so effective vulnerability assessment is very necessary for the Internet of things,and good vulnerability assessment can effectively classify the loopholes in the system,so that the work of vulnerability repair can be doubled with half the effort.From the point of view of the causes of the Internet of things vulnerabilities,most of them are caused by the logic defects or coding defects of the Internet of things source code.Therefore,before the official use of the Internet of things software,we should try to ensure its security and robustness.This paper studies the above two aspects to improve the security and stability of the Internet of things.In order to evaluate the vulnerability of the Internet of things system scientifically,this paper proposes a method of vulnerability related hazard assessment.Different from the traditional evaluation method which takes a single vulnerability as the evaluation unit,this paper takes the whole Internet of things system as an evaluation unit,analyzes the association relationship of vulnerabilities in the whole system by using vulnerability association graph,so as to find out the possible attack path,and then uses CVss V3.0 evaluation system is the basic index of calculation and introduces risk matrix.The calculation process includes not only the relationship between vulnerability and vulnerability,but also the attributes of vulnerability itself,so as to realize vulnerability Association evaluation.Experiments show that this method can effectively guide the security protection of the Internet of things.In view of the software defects of Internet of things software,this paper proposes a software defect prediction model.The model uses the improved local tangent space arrangement method to reduce the dimension,and the support vector machine is used as the classifier of the software defect prediction model.In order to reduce the accuracy of support vector machine caused by the lack of important information after dimension reduction,the traditional algorithm of dimension reduction based on LTSA uses fixed neighborhood,which leads to the loss of local structure and the sensitivity to noise.In this paper,an improved algorithm of local tangent space permutation(i LTSA)is proposed,which uses adaptive method to select neighborhood to ensure the integrity of local data structure,and at the same time uses robust method Type PCA decomposes neighborhood to achieve better robustness.The dimension is reduced by the improved LTSA algorithm,and SVM is trained and verified by the reduced dimension data.Compared with the prediction results of single SVM and LTSA-SVM,the prediction accuracy of i LTSA is higher,and the F-measure is improved by 3% ? 5%.