Label-Free Multivariate Time Series Anomaly Detection

异常检测 计算机科学 参数化复杂度 系列(地层学) 数据挖掘 异常(物理) 星团(航天器) 图形 多元统计 人工智能 模式识别(心理学) 机器学习 算法 理论计算机科学 古生物学 物理 生物 程序设计语言 凝聚态物理
作者
Qihang Zhou,Shibo He,Haoyu Liu,Jiming Chen,Wenchao Meng
出处
期刊:Cornell University - arXiv 被引量:1
标识
DOI:10.48550/arxiv.2312.11549
摘要

Anomaly detection in multivariate time series (MTS) has been widely studied in one-class classification (OCC) setting. The training samples in OCC are assumed to be normal, which is difficult to guarantee in practical situations. Such a case may degrade the performance of OCC-based anomaly detection methods which fit the training distribution as the normal distribution. In this paper, we propose MTGFlow, an unsupervised anomaly detection approach for MTS anomaly detection via dynamic Graph and entity-aware normalizing Flow. MTGFlow first estimates the density of the entire training samples and then identifies anomalous instances based on the density of the test samples within the fitted distribution. This relies on a widely accepted assumption that anomalous instances exhibit more sparse densities than normal ones, with no reliance on the clean training dataset. However, it is intractable to directly estimate the density due to complex dependencies among entities and their diverse inherent characteristics. To mitigate this, we utilize the graph structure learning model to learn interdependent and evolving relations among entities, which effectively captures complex and accurate distribution patterns of MTS. In addition, our approach incorporates the unique characteristics of individual entities by employing an entity-aware normalizing flow. This enables us to represent each entity as a parameterized normal distribution. Furthermore, considering that some entities present similar characteristics, we propose a cluster strategy that capitalizes on the commonalities of entities with similar characteristics, resulting in more precise and detailed density estimation. We refer to this cluster-aware extension as MTGFlow_cluster. Extensive experiments are conducted on six widely used benchmark datasets, in which MTGFlow and MTGFlow cluster demonstrate their superior detection performance.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
土豪的土豆完成签到 ,获得积分10
刚刚
刚刚
量子星尘发布了新的文献求助10
1秒前
科研小菜发布了新的文献求助10
1秒前
施耐德应助菠萝蜜采纳,获得10
2秒前
3秒前
5秒前
5秒前
紫鸢完成签到,获得积分10
5秒前
林卓琛发布了新的文献求助30
6秒前
8秒前
难过的达发布了新的文献求助10
9秒前
10秒前
12秒前
传奇3应助攀九年采纳,获得30
12秒前
liling发布了新的文献求助10
12秒前
打打应助土豆烧牛腩采纳,获得10
14秒前
深海鱼油DHA完成签到,获得积分10
15秒前
aqua_xin完成签到,获得积分0
16秒前
16秒前
机灵曼荷完成签到,获得积分10
17秒前
18秒前
田様应助难过的达采纳,获得10
18秒前
21秒前
21秒前
薄荷发布了新的文献求助10
21秒前
22秒前
23秒前
执着卿完成签到,获得积分10
23秒前
尤玉发布了新的文献求助10
23秒前
24秒前
谷雨秋发布了新的文献求助10
26秒前
科研通AI2S应助攀九年采纳,获得10
27秒前
Hello应助自然兰采纳,获得10
28秒前
YHY完成签到,获得积分10
28秒前
格格巫发布了新的文献求助10
28秒前
明亮的智宸完成签到,获得积分10
29秒前
30秒前
土豆烧牛腩完成签到,获得积分10
33秒前
CipherSage应助xiaotaiyang采纳,获得10
38秒前
高分求助中
【提示信息,请勿应助】请使用合适的网盘上传文件 10000
Continuum Thermodynamics and Material Modelling 2000
Chinesen in Europa – Europäer in China: Journalisten, Spione, Studenten 1200
Deutsche in China 1920-1950 1200
Electron microscopy study of magnesium hydride (MgH2) for Hydrogen Storage 800
Green Star Japan: Esperanto and the International Language Question, 1880–1945 800
Sentimental Republic: Chinese Intellectuals and the Maoist Past 800
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 物理 生物化学 纳米技术 计算机科学 化学工程 内科学 复合材料 物理化学 电极 遗传学 量子力学 基因 冶金 催化作用
热门帖子
关注 科研通微信公众号,转发送积分 3870718
求助须知:如何正确求助?哪些是违规求助? 3412820
关于积分的说明 10681363
捐赠科研通 3137252
什么是DOI,文献DOI怎么找? 1730812
邀请新用户注册赠送积分活动 834360
科研通“疑难数据库(出版商)”最低求助积分说明 781154