Research On Application Of K-means Algorithm And Logistic Regression Model In Network Security

The 21^st century is an era of information.The network utilization rate has reached the unprecedented height,while the application scope has expanded to all areas of life.Nowadays,the Internet is already an indispensable part to the development of social economy and national defense information.People will at a loss without the Internet.It is worth noting that while the Internet brings convenience to people's work and life,it is also prone to various types of security threats due to its characteristics of anonymity,openness,and decentralization.Therefore,Internet attacks are more serious and has a wider impact than before.There are a lot of information and information channels on the Internet.How to prevent various types of security threats effectively in this kind environment is a new requirement for network security in the new era.It requires not only the previous technical confrontation experience,but also need Big Data technology to cope with massive samples and give a fast and effective response.This is the intention of applying K-means algorithm and logistic regression model into network security.At present,the main defense technologies for network security threats include digital signature and CA authentication technology,firewall technology,network virus and prevention.This article will give the pros and cons of these parts through analysis and comparison.Focusing on the network virus and prevention,this paper expounds the process of virus analysis and the problems that are easy to encounter,explores how to applying big data technology to solve problems better,and gives solutions or ideas in combination with actual work.Use clustering algorithm and logistic regression model to cluster unknown samples and make fast and effective analysis for these samples.Anti-Virus vendors can receive thousands or even hundreds of millions of virus samples every day.How to deal with massive samples more quickly becomes a key issue.Most of the current treatment methods are identify the commonality of samples and make them a virus family,then start research work for this family.Based on the above situation,I have conducted in-depth research on the clustering algorithm,explored and summarized the algorithm suitable for clustering of massive samples,and made it has low complexity and high effectiveness.In addition,this paper implements a logistic regression model for predicting sample cluster results,which can automatically analyze the sample and output the predicted results.The model has a built-in credibility scoring system,and the high-confidence results can be directly used.This model can greatly reduce the complicated manual analysis process.Since that the application of big data technology in network security,especially in sample clustering and cluster result prediction,is relatively little,any research in this direction has certain theoretical and practical significance.After all,I also explored the emerging field of network security in the end of this paper,and gave the exploration and thinking of the application direction of big data technology.