异常检测模式中的持续学习研究

Research on continuous learning of anomaly detection pattern

提出了阶段化学习的概念,分阶段学习区分出了模式的完备性和不完备性。基于此概念,提出一种新的异常检测模式持续学习算法(PADPL)。仿真结果显示,PADPL能够满足由不完备性引起的异常检测模式持续学习的要求。

The concept of phase learning was proposed, and it could be used to determine whether a learning mode was complete or not. Then, a Phases Anomaly Detection Pattern Learning algorithm, PADPL, was put forward. Simulation shows that PADPL could meet the requirements of phases anomaly detection pattern learning which were caused by uncompleted learning mode.