|
| 一种分段集群异常作业预测方法 |
| A staged cluster anomaly job prediction method |
| 投稿时间：2018-12-23 修订日期：2019-01-22 |
| DOI： |
| 中文关键词: 集群异常作业 分阶段预测 实时预测 动态特征 门控递归单元 |
| 英文关键词: cluster abnormal job staged prediction real-time prediction dynamic features gated recurrent unit |
| 基金项目:国家自然科学基金民航联合研究(U1833107)；国家科技重大专项 (2012ZX03002002)；中央高校基本科研业务费 (ZYGX2018028)) |
|
| 摘要点击次数: 386 |
| 全文下载次数: 0 |
| 中文摘要: |
| 针对现有集群异常作业预测方法预测效率低、预测时间长的问题,提出一种分段集群异常作业预测方法(staged cluster anomaly job prediction method,SCAJP)。该方法分为离线预测和在线预测两个阶段；在离线阶段,依据作业子任务的静态特征对子任务终止状态进行预测,并仅在线预测此阶段的正常子任务所属作业；在线预测阶段中,在计算作业子任务的动态特征的同时,采用改进门控递归单元(improved gated recurrent unit,IGRU)神经网络根据动态特征实时预测任务终止状态是否异常。两个阶段的最后均根据作业与其子任务的相关性检索异常作业,实现对异常作业的预测。实验结果表明本文方法在预测灵敏度、精确度和预测时间方面明显优于其它方法。 |
| 英文摘要: |
| Aiming at the problem of low prediction efficiency and long prediction time of the existing cluster anomaly job prediction methods, a staged cluster anomaly job prediction method (SCAJP) was proposed in this paper. This method was divided into offline stage and online stage. The final state of the job’s sub-tasks were predicted according to its’ static features in the offline stage, then only did online prediction for the job to which the normal sub-task belonged; In online stage, while calculating the dynamic features of the job’s sub-tasks, the improved gated recurrent unit (IGRU) neural network was used to predict whether the task termination status was abnormal according to the dynamic features in real time. At the end of the both stages, the abnormal job was obtained based on the relevance between the job and its sub-tasks to finish the prediction of the anomaly job. The experimental results show that the proposed method outperforms other methods in terms of prediction sensitivity, accuracy and prediction time obviously. |
|
View Fulltext
查看/发表评论 下载PDF阅读器 |
| 关闭 |