【摘要】 基于支持向量机提出一种方法,能够从历史告警中挖掘出告警的预测知识。首先把告警预测问题转化成机器学习问题,然后通过在告警序列上滑动窗口形成样本去训练支持向量机,从而建立对某种特定告警的预测模型。实验验证了该方法的有效性,并考察了一些关键参数对于预测性能的影响。更多还原

【Abstract】 In this paper a method based on support vector machine is presented, which can discover prediction knowledge from historical alarms. Firstly the prediction problem is tranformed to a machine learning problem, then samples are formed by sliding windows on alarm sequence to train svm, thus the prediction model for a given alarm is established. In the experiment the validity of the method is proved and impact on the prediction performance about some key parameters is analyzed.更多还原