[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123075-en":3,"doc-seo-123075-105":30,"detail-sidebar-cat-0-en-105":91},{"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":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},123075,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",6,"Technology","Machine Learning in High Volume Media Manufacturing - Deployment and Adaptive Failure Detection","Errors or failures in high-volume manufacturing can create major losses in both time and money. Early identification has traditionally relied on rule-based algorithms, but these approaches are slow to apply, difficult to maintain, and poorly adaptable to changing designs and day-to-day behavioral variations. This work presents a scalable program that combines rule-based decisions with machine learning models to learn, adapt, and detect issues efficiently across large unit counts. The system is deployed at scale using modern production technologies.","Machine Learning in High Volume Media Manufacturing  \nSiddarth Reddy Karuka1*, Abhinav Sunderrajan2*, Zheng Zheng3\\#, Yong Woon Tiean3, Ganesh Nagappan2*, and Allan Luk4*  \n1 – Seagate Technology, Bloomington, Minnesota, USA  \n2 – Seagate Technology, Shugart, Singapore  \n3 – Seagate Technology, Woodlands, Singapore  \n4 – Seagate Technology, Longmont, Colorado, USA  \n* These authors are not currently at Seagate  \n\\# Corresponding author ([zheng.zheng@seagate.com](zheng.zheng@seagate.com))  \nAbstract: Errors or failures in a high-volume manufacturing environment can have significant impact that can result in both the loss of time and money. Identifying such failures early has been a top priority for manufacturing industries and various rule-based algorithms have been developed over the years. However, catching these failures is time consuming and such algorithms cannot adapt well to changes in designs, and sometimes variations in everyday behavior. More importantly, the number of units to monitor in a high-volume manufacturing environment is too big for manual monitoring or for a simple program. Here we develop a novel program that combines both rule-based decisions and machine learning models that can not only learn and adapt to such day-to-day variations or long-term design changes, but also can be applied at scale to the high number of manufacturing units in use today. Using the current state-of-theart technologies, we then deploy this program at-scale to handle the needs of ever-increasing demand from the manufacturing environment.  \nKeywords: Manufacturing; Machine Learning, DBSCAN, Deployment, Kubernetes.  \n1. Introduction  \nHigh vacuum integrity and thin film quality are essential to producing world-class quality and reliable media through multi-layer processing in high throughput sputter process. The sputter process involves a disk moving through a sputter machine that consists of multiple stations, each with their own purpose (Figure 1) . A disk carrier carries the disks across the stations such that any given individual disk goes through all the sputter stations sequentially. The stations, depending on the process performed in each individual station, are maintained at a very low pressure or high vacuum. Slot valves separate these chambers from the outside environment and thus help in maintaining the high vacuum. Any degradation or faults in these slot valves lead to small openings that can cause unwanted air to rush into the station chambers, referred to as a leak here after. Such leaks not only lead to wastage of raw material but also affect the quality of the disk, thereby affecting the quality of downline products. These leaks show up in the pressure cycles, where each cycle is defined as the pressure profile observed for each individual station between the time a slot valve is closed and opened again. In other words, a cycle is defined by the time a disk spends undergoing a specific process in a particular station, before moving on to the next  \nstation. Catching such minor leak cycles is usually a daunting process and is described in detail in the following sections.  \nFigure 1: An example sputter machine (details and specifics modified or withheld intentionally)  \nSince each individual station has their own process or function associated with it, the pressure profile looks different for each of them. However, in general, the stations can be attributed to two types: process and non-process stations. Although the pressure profiles arising from all process and all nonprocess stations look similar, differences between stations are sufficient to confuse a generic leak detection algorithm. Thus, it is important to have a program that can consider pressure profiles that are characteristic to each individual station. This brings out an important problem the problem of scale. With hundreds of machines on the factory floor, and tens of stations per machine, there are thousands of stations overall. Thus, it is quite diffic","cbCaie0aqyz2knAL","https://ap.wps.com/l/cbCaie0aqyz2knAL","pdf",1061691,1,13,"English","en",105,"# Introduction\n## Manufacturing scale and leak detection challenges\n# Materials and Methods\n## Supervised vs. unsupervised learning\n## Scalable program architecture (rules + ML)","[{\"question\":\"Why is leak detection in high-volume media manufacturing challenging at scale?\",\"answer\":\"Each sputter station has distinct pressure-cycle behavior, so generic leak detection algorithms are easily confused. With many machines and stations overall, manual rule development and deployment becomes impractical.\"},{\"question\":\"How does the proposed approach differ from traditional rule-based algorithms?\",\"answer\":\"It combines rule-based decisions with machine learning models that can learn and adapt to both short-term variations and long-term design changes, enabling more flexible monitoring.\"},{\"question\":\"Why were unsupervised machine learning models chosen?\",\"answer\":\"Labeled leak profiles are limited and may not include new or previously unknown leak patterns. Changes in machines or sputter design can also render supervised models ineffective, requiring costly retraining.\"}]","Machine Learning in High Volume Media Manufacturing - Deployment and Adaptive Failure Detection | PDF",1785814519,33,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"machine-learning-in-high-volume-media-manufacturing-deployment-and-adaptive-failure-detection","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/technology/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/machine-learning-in-high-volume-media-manufacturing-deployment-and-adaptive-failure-detection/123075/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is leak detection in high-volume media manufacturing challenging at scale?","Question",{"text":75,"@type":76},"Each sputter station has distinct pressure-cycle behavior, so generic leak detection algorithms are easily confused. With many machines and stations overall, manual rule development and deployment becomes impractical.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the proposed approach differ from traditional rule-based algorithms?",{"text":80,"@type":76},"It combines rule-based decisions with machine learning models that can learn and adapt to both short-term variations and long-term design changes, enabling more flexible monitoring.",{"name":82,"@type":73,"acceptedAnswer":83},"Why were unsupervised machine learning models chosen?",{"text":84,"@type":76},"Labeled leak profiles are limited and may not include new or previously unknown leak patterns. Changes in machines or sputter design can also render supervised models ineffective, requiring costly retraining.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,113,118,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":111,"slug":112},50,"technology",{"id":114,"doc_module":4,"doc_module_name":46,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":46,"category_name":120,"show_sort_weight":121,"slug":122},8,"Research & Report",30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]