The Mobile Data Security Detection Technology Research Based On Machine Learning

With the rapid development of mobile Internet, the security problem is becoming more and more serious. As malwares’attack methods vary day by day, the traditional security detection methods such as signature scanning have been unable to handle the growing types of the malicious attacks. Security detection methods based on machine learning usually have the ability to detect unknown attacks. It can effectively make up the disadvantages of traditional security detection methods, it’s very significant to study it.This paper analyzes the existing security detection methods and their limitations. Considering many malicious software have characteristics of network behavior, this paper proposes to detect the mobile network data to ensure the security of the mobile Internet. This paper first proposed a security detection model based on classification algorithms to detection the HTTP request data; then verified the model’s validity. The following is the main work:1. Analyze the characteristics of the HTTP request and its malicious features, propose to construct the features from the respect of structure, length, character and keywords, design a lot of features. Select effective features based on information gain feature selection algorithms, use the selected features to vectorised data.2. Build classification models based on the classification algorithms, which can classify the HTTP request into normal class or anomalous class. Build the classification model respectively on Naive Bayes algorithm, C4.5decision tree algorithm and SVM algorithms. The experiment results shows that the designed features can effectively represent the data and the classification models based C4.5decision tree and SVM works well.3. Implement an intrusion detection system based on C4.5decision tree algorithm, which integrates the functions of data preprocessing, model training, data detection and alert. Use it as a module of the safety detection system to realize an initiative security defense of the network data.