[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-127391-en":3,"doc-seo-127391-105":31,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},127391,962085571259,"Theodora","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",6,"Technology","OML-AD - Online Machine Learning for Anomaly Detection in Time Series Data","Time series are ubiquitous across domains such as manufacturing sensors, financial streams, and climate monitoring, and they require reliable handling of regression, classification, and segmentation tasks. Accurate solutions depend on filtering abnormal observations that deviate from typical temporal behavior. Many existing methods assume stationary or independent data and fail under non-stationarity. OML-AD introduces an online machine learning approach that detects anomalies while adapting to concept drift, and provides an implementation in the River Python library with improved accuracy and computational efficiency.","arXiv :2409 .09742v1 [ cs .LG] 15 Sep 2024  \nOML-AD: ONLINE MACHINE LEARNING FOR ANOMALY DETECTION IN TIME SERIES DATA  \nSebastian Wette  \nDepartment of Computer Science Technische Universitt Darmstadt Darmstadt, 64289, Germany  \n[sebastian.wette@stud.tu-darmstadt.de](sebastian.wette@stud.tu-darmstadt.de)  \nFlorian Heinrichs  \nDepartment of Medical Engineering and Technomathematics  \nFH Aachen-University of Applied Sciences J¨ulich, 52428, Germany [f.heinrichs@fh-aachen.de](f.heinrichs@fh-aachen.de)  \nABSTRACT  \nTime series are ubiquitous and occur naturally in a variety of applications – from data recorded by sensors in manufacturing processes, over financial data streams to climate data. Different tasks arise, such as regression, classification or segmentation of the time series. However, to reliably solve these challenges, it is important to filter out abnormal observations that deviate from the usual behavior of the time series. While many anomaly detection methods exist for independent data and stationary time series, these methods are not applicable to non-stationary time series. To allow for non-stationarity in the data, while simultaneously detecting anomalies, we propose OML-AD, a novel approach for anomaly detection (AD) based on online machine learning (OML) . We provide an implementation of OML-AD within the Python library River and show that it outperforms state-ofthe-art baseline methods in terms of accuracy and computational efficiency.  \n1 INTRODUCTION  \nToday’s technology ecosystems often rely on anomaly detection for monitoring and fault detection (Ahmad et al., 2017) . There are various approaches to anomaly detection (Aggarwal, 2017), but machine-learning (ML) based methods stand out as the most used in real-world use cases (Laptev et al., 2015) . Their ability to efficiently process and learn from large datasets led to widespread adoption. However, the general use of classical ML algorithms trained on large batches of data needs to be revised to work for today’s dynamically changing and fast-paced systems. The primary concern is the phenomenon of concept drift, which occurs when the statistical properties of the predicted target variable change over time (Lu et al., 2018) . As a result, models trained on historical data batches may become outdated, and performance can deteriorate when forecasting (Lu et al., 2018) because of their inability to adapt to changes in the data (Chatfield, 2000) . Anomaly detection techniques that rely on accurate predictions of an underlying model suffer from this phenomenon especially. Different approaches to handling concept drift have been proposed in the past (Gama et al., 2014; Lu et al., 2018) . One approach is to retrain the model once a change point is detected. While approaches like this can produce satisfactory results, they are complex and costly. Further, they might not detect smooth changes, as occurring in many real-world settings. Hence, there is a need for a robust and dynamic anomaly detection solution that is cheap, performant and able to work with gradual changes.  \nIn this context, online ML emerges as a potential solution. Unlike their batch-learning counterparts, online learning algorithms incrementally perform optimization steps in response to new concepts’influence in the data (Shalev-Shwartz et al., 2012) . This continuous learning paradigm enables  \nthese algorithms to adapt to changing distributions in data without retraining, thereby ensuring the model’s sustained precision. We aim to leverage the features of online learning for predictive anomaly detection on time series data under concept drift to counter common problems of batchtrained ML models.  \nWe propose to combine the existing ideas of prediction-based anomaly detection with online machine learning to create a more dynamic and robust solution.  \nTo compare the proposed approach to similar prediction-based anomaly detection methods commonly employed (e.g., Meta’s Prophet Taylor and Letham, 20","cbCaitgVEnpNoTzs","https://ap.wps.com/l/cbCaitgVEnpNoTzs","pdf",641658,2,1,14,"English","en",105,"# Abstract\n# Introduction\n# Related Work","[{\"question\":\"Why do classical anomaly detection methods struggle with time series concept drift?\",\"answer\":\"Many approaches assume stationary data or rely on batch-trained models, which can become outdated when statistical properties change over time. This can degrade forecasting and anomaly detection performance.\"},{\"question\":\"What problem does OML-AD address?\",\"answer\":\"OML-AD targets non-stationary time series by combining prediction-based anomaly detection with online machine learning, enabling adaptation while detecting anomalies under concept drift.\"},{\"question\":\"How is OML-AD implemented and evaluated?\",\"answer\":\"The work provides an OML-AD implementation using the River Python library and compares it against state-of-the-art baselines using experiments on synthetic and real time series.\"}]","OML-AD - Online Machine Learning for Anomaly Detection in Time Series Data | PDF",1785938646,35,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":29},"oml-ad-online-machine-learning-for-anomaly-detection-in-time-series-data","",{"@graph":37,"@context":86},[38,54,69],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,48,51],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":20},"https://docshare.wps.com/document/","Document",{"item":49,"name":12,"@type":44,"position":50},"https://docshare.wps.com/document/technology/",3,{"item":52,"name":13,"@type":44,"position":53},"https://docshare.wps.com/document/oml-ad-online-machine-learning-for-anomaly-detection-in-time-series-data/127391/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":42,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-23","2026-08-05",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"Why do classical anomaly detection methods struggle with time series concept drift?","Question",{"text":76,"@type":77},"Many approaches assume stationary data or rely on batch-trained models, which can become outdated when statistical properties change over time. This can degrade forecasting and anomaly detection performance.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"What problem does OML-AD address?",{"text":81,"@type":77},"OML-AD targets non-stationary time series by combining prediction-based anomaly detection with online machine learning, enabling adaptation while detecting anomalies under concept drift.",{"name":83,"@type":74,"acceptedAnswer":84},"How is OML-AD implemented and evaluated?",{"text":85,"@type":77},"The work provides an OML-AD implementation using the River Python library and compares it against state-of-the-art baselines using experiments on synthetic and real time series.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,111,114,119,124,129,132,136],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":47,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":47,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":112,"slug":113},50,"technology",{"id":115,"doc_module":4,"doc_module_name":47,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":120,"doc_module":4,"doc_module_name":47,"category_name":121,"show_sort_weight":122,"slug":123},8,"Research & Report",30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":47,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":47,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]