[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-341718-105":59,"doc-detail-341718-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","towards-practical-learned-indexing-extended-abstracts","Towards Practical Learned Indexing - Extended Abstracts","","Latest research explores learned indexes as replacements for conventional index structures, yet current methods often rely on many hyperparameters, lack error guarantees, and remain costly to build. Practical Learned Index (PLEX) introduces a learned indexing design with only a single hyperparameter, the maximum prediction error. PLEX improves the build–lookup trade-off and provides an explicit error bound by combining a spline layer with a multi-level radix layer, then tuning the radix structure via an analysis of spline-point distributions.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/towards-practical-learned-indexing-extended-abstracts/341718/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/towards-practical-learned-indexing-extended-abstracts/341718.png","ImageObject",300,407,{"name":92,"@type":93},"Theodora","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-04","2026-09-22",true,{"@type":102,"interactionType":103,"userInteractionCount":39},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What is PLEX and what problem does it aim to solve?","Question",{"text":112,"@type":113},"PLEX is a Practical Learned Index that targets shortcomings of prior learned indexes, especially the large number of hyperparameters, missing error guarantees, and expensive build costs.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How many hyperparameters does PLEX require?",{"text":117,"@type":113},"PLEX uses only a single hyperparameter: the maximum prediction error (maximum allowed error bound).",{"name":119,"@type":110,"acceptedAnswer":120},"What are the main components of PLEX?",{"text":121,"@type":113},"PLEX combines an error-bounded spline model with a radix-based indexing layer implemented via CHT, and it builds the spline first before indexing and tuning the radix layer.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},341718,1790822017,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":39,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":24,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":143,"read_time":144,"preview_page_count":19,"preview_urls":145},687197207919,"https://ap-avatar.wpscdn.com/avatar/a000253d6f5f7c60be?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779446848396160552","Towards Practical Learned Indexing (Extended Abstracts)  \nMihail Stoian  \nTUM  \n[mihail.stoian@tum.de](mihail.stoian@tum.de)  \nAndreas Kipf  \nMIT CSAIL [kipf@mit.edu](kipf@mit.edu)  \nRyan Marcus MIT CSAIL, Intel Labs [ryanmarcus@mit.edu](ryanmarcus@mit.edu)  \nTim Kraska  \nMIT CSAIL [kraska@mit.edu](kraska@mit.edu)  \narXiv :2108 .05117v2 [ cs .DB] 6 Nov 2021  \nABSTRACT  \nLatest research proposes to replace existing index structures with learned models. However, current learned indexes tend to have many hyperparameters, often do not provide any error guarantees, and are expensive to build. We introduce Practical Learned Index (PLEX) . PLEX only has a single hyperparameter 􀀏 (maximum prediction error) and o􀀋ers abetter trade-o􀀋 between build and lookup time than stateof-the-art approaches. Similar to RadixSpline, PLEX consists of a spline and a (multi-level) radix layer. It 􀀌rst buildsa spline satisfying the given 􀀏 and then performs an ad-hoc analysis of the distribution of spline points to quickly tune the radix layer.  \nAIDB Workshop Reference Format:  \nMihail Stoian, Andreas Kipf, Ryan Marcus, Tim Kraska. Towards Practical Learned Indexing. AIDB 2021.  \n1. INTRODUCTION  \nWe introduce Practical Learned Index (PLEX) . Compared to existing learned indexes, PLEX only has a single hyperparameter 􀀏 (maximum prediction error) and is hence easy to use. PLEX builds upon RadixSpline [7] and CHT [2], and uses a spline layer to ensure an 􀀏 error bound and a radix layer that is inspired by CHT.  \nRadixSpline. RadixSpline (RS) [7] consists of a linear spline model that approximates the CDF of the data within an error bound and a radix table that indexes the computed spline points. The main feature of RS is that it can be built with a constant amount of work per new element, which allows a single-pass build phase. However, when the radix table is not able to properly index the spline keys (i.e., in the presence of outliers), RS can have poor performance.  \nPLEX addresses two problems of RS: (i) the radix layer is hard to parametrize and (ii) the radix layer can be a􀀋ected by outliers. Despite the fact that PLEX needs to perform an additional pass over the spline points array, PLEX manages to retain the high build performance of RS.  \nThis article is published under a Creative Commons Attribution License ([http://creativecommons.org/licenses/by/3.0/](http://creativecommons.org/licenses/by/3.0/)), which permits distribution and reproduction in any medium as well allowing derivative works, provided that you attribute the original work to the author(s) and AIDB 2021 . 3rd International Workshop on Applied AI for Database Systems and Appli cations (AIDB'21), August 20, 2021, Copenhagen, Denmark.  \nHist-Tree. Hist-Tree (HT) [2] approximates the data distributed as a histogram, instead of using functions, with a fast radix tree-based traversal method to 􀀌nd the right histogram bucket. As such, HT has similarities with a traditional radix tree, as its indexing approach also uses the binary representation of the keys, and with a learned index as it tries to approximate the data distribution rather using the traditional key-comparison present in B-tree traversals. HT recursively splits the data into bins until a certain threshold for the number of elements is reached. Apart from allowing updates, Hist-Tree also comes in a compact version (CHT), optimized for read-only workloads. CHT is essentially a lookup table which stores the nodes of the initial HT and their pre􀀌x sums. CHT can either be built from a HT, by running a depth-􀀌rst pre-order tree traversal, or, as we have implemented it in this work, directly from the data itself. The latter is more advantageous for our case, as keys can be processed in chunks. While CHT did not have an auto-tuner, we now introduce one as part of this work (auto-tuning CHT is part of our auto-tuning process) . Other Related Work. The 􀀌rst learned index, which paved the way for the development of new alternatives of i","cbCaij4ZzNedkCQf","https://ap.wps.com/l/cbCaij4ZzNedkCQf","pdf",1395027,"English","# Abstract\n# Introduction\n## RadixSpline\n## Hist-Tree (HT) and CHT\n# PLEX\n## Build Spline\n## Index Spline Points","[{\"question\":\"What is PLEX and what problem does it aim to solve?\",\"answer\":\"PLEX is a Practical Learned Index that targets shortcomings of prior learned indexes, especially the large number of hyperparameters, missing error guarantees, and expensive build costs.\"},{\"question\":\"How many hyperparameters does PLEX require?\",\"answer\":\"PLEX uses only a single hyperparameter: the maximum prediction error (maximum allowed error bound).\"},{\"question\":\"What are the main components of PLEX?\",\"answer\":\"PLEX combines an error-bounded spline model with a radix-based indexing layer implemented via CHT, and it builds the spline first before indexing and tuning the radix layer.\"}]","Towards Practical Learned Indexing - Extended Abstracts | PDF",1790043353,13,[146,150,154,158],{"page_no":8,"url":147,"width":148,"height":149},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/8bceff25e4913f3660f23b9bc8432221.webp",983,1272,{"page_no":14,"url":151,"width":152,"height":153},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/9b94f35f9a1e14b0f5e28044512e3edf.webp",982,1270,{"page_no":81,"url":155,"width":156,"height":157},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/23a0d04d715168fc183dd1a59c214bf0.webp",981,1269,{"page_no":19,"url":159,"width":160,"height":161},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/eae3277332f44c99ad50aa1ec60f5671.webp",980,1268]