摘要
本文针对现有恶意软件检测系统无法保证数据库恶意行为检测的效率和精度的问题,设计了一个基于机器学习中主动学习原理的数据库恶意行为检测方法并在MySQL上实现了原型系统。测试表明该系统对数据库恶意行为检测具有较高的检测率,较低的误报率和漏报率。
First,the conception of malicious behavior characteristics signatures from the database session behavior is defined.Then,the risk factor to describe the dangers of the malicious behavior of a short sequence is proposed.Last,the risk rand is introduced to divide the software into malicious software and normal software.And a prototype system is developed in MySQL.The experimental results show that the malicious behavior detection correct rate of about 82% with this method which has a high detection correct rate and a low false alarm rate and false negative rate.
出处
《网络安全技术与应用》
2012年第10期62-64,共3页
Network Security Technology & Application
关键词
数据库管理系统
恶意行为检测
机器学习
MYSQL
Databse security
Malware detection
Intrusion detection
Behavior signature
Machine Learning
Active Learning