Web Access Log Anomaly Detection Based on Deep Learning
Web Access Log Anomaly Detection Based on Deep Learning
复制标题
基于深度学习的Web访问日志异常检测
DOI：
10.1145/3451471.3451491
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Quan Liu
中科院分区：
文献类型：
--
作者：
Quan Liu
网络信息安全正在成为许多拥有大数据的企业和研究机构的重要课题。通过网络的信息泄露正在迅速增加。许多企业和研究机构的管理者还没有认识到信息安全的重要性。本文描述了异常检测的研究。我们测量原始网络访问日志的格式和内容。目的是分析每个用户的访问类型和访问类型分布，并将其转化为机器可以学习的类型。通过转换后的日志，可以建立神经网络模型来学习如何检测异常访问日志并向相关计算机管理员发出警报。主要目标是设计一个可以协助计算机管理员完成异常检测工作的模型。
Network information security is becoming a vital topic for many companies and research institutes with big data. Information leakage via cyber is rising rapidly. Many companies and research institutes managers have not realized the significance of information security. This paper describes an investigation of anomaly detection. We measure the format and content of the original web access log. The intent is to analyze types of access and distribution of access types of each user and to transform them into the type that can be learned by the machine. With the transformed log, a neural network model can be put in place to learn how to detect an abnormal access log and alert the relative computer managers. The main goal is to design a model that can assist computer managers to do the anomaly detection job.