arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.03324cs.LGcs.AI
收藏

基于重构型状态空间模型的时间序列异常检测

Time series anomaly detection with reconstruction-based state-space models

Novo Nordisk A/S(诺和诺德公司);China Academy of Railway Sciences(中国铁道科学研究院);China Railway Test & Certification Center Limited(中铁检验认证中心有限公司)

Fan Wang, Keli Wang, Boyu Yao

AI总结:

针对多变量时间序列异常检测难题,提出一种结合LSTM编解码器、双向状态转移建模与马氏距离评估的无监督方法,在合成及真实数据集上取得更优检测性能。

摘要:

数字化的最新进展使得多变量时间序列数据在各领域均可获取,从而支持对运营的实时监控。在这些场景中识别异常数据模式并检测潜在故障十分重要,但也极具挑战性。本文提出了一种新型无监督时间序列异常检测方法。该框架联合学习观测模型与动态模型,并基于正常样本估计模型不确定性。具体而言,采用基于长短期记忆网络(LSTM)的编码器-解码器来表征观测空间与潜空间之间的映射关系,同时利用反向与前向时序信息对状态的双向转移进行建模。潜空间正则化对正常样本的状态施加约束,并使用马氏距离评估异常程度。在合成数据集与真实数据集上的实证研究表明,该方法在异常检测任务中具备优异性能。

英文摘要:

Recent advances in digitization have led to the availability of multivariate time series data in various domains, enabling real-time monitoring of operations. Identifying abnormal data patterns and detecting potential failures in these scenarios are important yet rather challenging. In this work, we propose a novel unsupervised anomaly detection method for time series data. The proposed framework jointly learns the observation model and the dynamic model, and model uncertainty is estimated from normal samples. Specifically, a long short-term memory (LSTM)-based encoder-decoder is adopted to represent the mapping between the observation space and the latent space. Bidirectional transitions of states are simultaneously modeled by leveraging backward and forward temporal information. Regularization of the latent space places constraints on the states of normal samples, and Mahalanobis distance is used to evaluate the abnormality level. Empirical studies on synthetic and real-world datasets demonstrate the superior performance of the proposed method in anomaly detection tasks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑