类图像处理面向大数据XSS入侵智能检测研究  

SIMILAR IMAGES PROCESSING FOR BIG DATA XSS INTRUSION INTELLIGENT DETECTION

作　　者：张海军[1] 陈映辉[2] Zhang Haijun;Chen Yinghui(School of Computing,Jiaying University,Meizhou 514015,Guangdong,China;School of Mathematics,Jiaying University,Meizhou 514015,Guangdong,China)

机构地区：[1]嘉应学院计算机学院,广东梅州514015 [2]嘉应学院数学学院,广东梅州514015

出　　处：《计算机应用与软件》2020年第2期288-294,共7页Computer Applications and Software

基　　金：国家自然科学基金项目(61171141,61573145);广东省自然科学基金重点项目(2014B010104001,2015A030308018);广东省普通高等学校人文社会科学省市共建重点研究基地项目(18KYKT11);广东省嘉应学院自然科学基金重点项目(2017KJZ02)。

摘　　要：通过类图像处理方法对访问流量语料库大数据进行词向量化处理,实现面向大数据XSS入侵智能检测研究。利用类图像处理方法进行数据获取、数据清洗、数据抽样、特征提取等数据预处理;设计基于神经网络的词向量化算法,实现词向量化得到词向量大数据;通过理论分析和推导,实现多种不同深度的深层神经网络智能检测算法;设计不同的超参数并进行反复的实验,分别得到最高识别率、最低识别率、识别率均值、方差、标准差、识别率变化曲线图和平均绝对误差变化曲线图等结果。实验结果表明,该方法具有识别率高、稳定性好、总体性能优良等特点。In this paper,we carry out word vectorization processing on the big data of the access traffic corpus by class image processing method,and realize the XSS intrusion intelligent detection for big data.We used similar images processing methods for data acquisition,data cleaning,data sampling,feature extraction and other data preprocessing.Then,the neural network-based word vectorization algorithm was designed to realize the word vectorization and obtain the word vector big data.Through theoretical analysis and deductions,a variety of deep neural network intelligent detection algorithms with different depths were realized.Finally,we designed different hyperparameters and carried out repeated experiments to obtain the results of the highest recognition rate,the lowest recognition rate,the recognition rate mean,the variance,the standard deviation,the recognition rate change curve and the change curve of average absolute error,etc.As a result,it is proved that our method has the characteristics of high recognition rate,good stability and excellent overall performance.

关 键 词：Web入侵检测 XSS 自然语言处理 深度学习 深度神经网络 大数据 网络安全 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]