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 2405.11238cs.LGcs.AI
收藏

SimAD:一种简单的基于不相似性的时间序列异常检测方法

SimAD: A Simple Dissimilarity-based Approach for Time Series Anomaly Detection

Pengcheng Laboratory, Shenzhen, Guangdong, 518066, China(鹏城实验室,深圳,广东,518066,中国);School of Future Technology, South China University of Technology, Guangzhou, Guangdong 510650, China(未来技术学院,华南理工大学,广州,广东 510650,中国);School of Computer Science and Engineering, South China University of Technology, Guangzhou, Guangdong 510650, China(计算机科学与工程学院,华南理工大学,广州,广东 510650,中国);Department of Information and Computing Science, Chongqing Jiaotong University(信息与计算科学系,重庆交通大学);Department of Computer Science at the University of Oxford(牛津大学计算机科学系);Alibaba Group(阿里巴巴集团);Department of Industrial and Systems Engineering in The Hong Kong Polytechnic University, China(香港理工大学工业与系统工程系,中国)

Zhijie Zhong, Zhiwen Yu, Xing Xi, Yue Xu, Wenming Cao, Yiyuan Yang, Kaixiang Yang, Jane You

AI总结:

针对时间序列异常检测中时间上下文有限和评估指标缺陷等问题,提出基于不相似性的SimAD方法,结合分块特征提取、ContrastFusion模块及新评估指标,在多个数据集上显著优于现有方法。

摘要:

尽管基于重构的深度学习方法很流行,但时间序列异常检测仍然是一个巨大的挑战。现有方法常常在有限的时间上下文、正常模式表示不足以及有缺陷的评估指标方面遇到困难,这些都阻碍了它们在检测异常行为方面的有效性。为了解决这些问题,我们引入了一种简单的基于不相似性的时间序列异常检测方法,称为SimAD。具体而言,SimAD首先整合了一个基于分块的特征提取器,能够处理扩展的时间窗口,并采用EmbedPatch编码器来充分整合正常行为模式。其次,我们在SimAD中设计了一个创新的ContrastFusion模块,通过突出正常和异常数据之间的分布差异来增强异常检测的鲁棒性。第三,我们引入了两个鲁棒的增强评估指标,即无偏隶属度(UAff)和归一化隶属度(NAff),旨在通过提供更好的区分性和语义清晰度来克服现有指标的局限性。理论和实验分析都证明了这两个指标的可靠性。在七个不同的时间序列数据集上进行的实验清楚地表明,与最先进的方法相比,SimAD具有优越的性能,在六个多变量数据集上,F1相对提高了19.85%,Aff-F1提高了4.44%,NAff-F1提高了77.79%,AUC提高了9.69%。代码和预训练模型可在https://github.com/EmorZz1G/SimAD获取。

英文摘要:

Despite the prevalence of reconstruction-based deep learning methods, time series anomaly detection remains a tremendous challenge. Existing approaches often struggle with limited temporal contexts, insufficient representation of normal patterns, and flawed evaluation metrics, all of which hinder their effectiveness in detecting anomalous behavior. To address these issues, we introduce a $\textbf{Sim}$ple dissimilarity-based approach for time series $\textbf{A}$nomaly $\textbf{D}$etection, referred to as $\textbf{SimAD}$. Specifically, SimAD first incorporates a patching-based feature extractor capable of processing extended temporal windows and employs the EmbedPatch encoder to fully integrate normal behavioral patterns. Second, we design an innovative ContrastFusion module in SimAD, which strengthens the robustness of anomaly detection by highlighting the distributional differences between normal and abnormal data. Third, we introduce two robust enhanced evaluation metrics, Unbiased Affiliation (UAff) and Normalized Affiliation (NAff), designed to overcome the limitations of existing metrics by providing better distinctiveness and semantic clarity. The reliability of these two metrics has been demonstrated by both theoretical and experimental analyses. Experiments conducted on seven diverse time series datasets clearly demonstrate SimAD's superior performance compared to state-of-the-art methods, achieving relative improvements of $\textbf{19.85%}$ on F1, $\textbf{4.44%}$ on Aff-F1, $\textbf{77.79%}$ on NAff-F1, and $\textbf{9.69%}$ on AUC on six multivariate datasets. Code and pre-trained models are available at https://github.com/EmorZz1G/SimAD.

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。

↑