Research On Information Security Model Based On PCA And BP Neural Network

With the rapid development of computer information technology,the problem of information security is becoming more and more serious.Information technology has been closely linked with our life,work and even national security,but the severe form of information security is a problem we can not ignore.At present,the information security defense of computer system mainly relies on network security devices such as firewalls,network gates,etc.The main defense mode of these network security devices is to protect the information security by means of a comparative identification and killing defense method with the help of the existing virus feature library.There is a lagging defect in this defense method of comparing virus banks.When a new type of virus appears,it can not immediately detect and send feedback.Although traditional information security technologies such as firewalls and encryption have certain defensive effects,they belong to the category of static security technology.Such security devices can not resist new viruses or attacks that do not have such abnormal data characteristics.In the face of such viruses or attacks,if there is no corresponding security strategy,security devices(such as traditional firewalls and network gates).It loses the function of filtering blocking,which makes network system leak dangerous,and attacks on legitimately open ports mostly fail to play a role of security protection.With the continuous development of information technology,a large number of new Trojan horse viruses and attack means have appeared.Traditional security methods based on feature library,security strategy and port control,facing new viruses and attack means,can not defend against computer viruses attack in a way of pre-active identification,and can not fully meet the requirements of current network security.Some machine learning algorithms in artificial intelligence,such as neural network algorithm,genetic algorithm and fuzzy technology,have the characteristics of autonomous learning,which can achieve an active security defense effect in the field of information security.By means of dynamic defense,while protecting the network to detect and block abnormal data in time,it can also provide real information system virus attacks.Protection of time.Therefore,the application of machine learning algorithm to information security has become a research hotspot in the field of information security.Because the anomalous data in information security is usually high-dimensional and massive,traditional detection methods are inefficient for anomalous data detection and low recognition rate for unknown anomalous data.This paper presents an information security detection method based on PCA and BP neural network.After training and learning the known abnormal data,this method can not only recognize the known abnormal data efficiently,but also recognize the abnormal data that has not been trained and learned.This BP neural network anomaly data recognition method based on PCA is applied to information security defense,which can solve the defect of traditional anomaly data recognition method that can not be proactively defended in advance.Finally,the simulation results under the environment of MATLAB show that:(1)The information security model which combines the PCA of statistical method and the BP neural network of machine learning algorithm in the field of information security not only improves the recognition rate of the known abnormal data,but also has a good recognition effect on the unknown and unknown new abnormal data.(2)The improved BP neural network algorithm accelerates the convergence speed of the algorithm,improves the efficiency of the neural network,and improves the recognition rate of network abnormal data.(3)The experimental results of simulation and comparison PCA-improved BP neural network detection model has higher recognition rate and lower false alarm rate than the traditional three classical algorithms.