[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-124309-en":3,"doc-seo-124309-105":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},124309,1374391974468,"Eden","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","Active learning of molecular data for task-specific objectives - Research article","Active learning (AL) offers data-efficient machine learning for property prediction, but computational savings depend strongly on the specific application. A systematic assessment evaluates AL on three molecular datasets across two scientific tasks: building compact informative datasets and performing targeted molecular searches. Using Gaussian processes with many-body tensor representations, the study compares acquisition strategies, batch sizes, and GP noise settings. Results show strong task-dependent behavior, including up to 64% data savings for targeted searches.","RESEARCH ARTICLE | JANUARY 02 2025  \nActive learning of molecular data for task-specific objectives  \nKunal Ghosh  ; Milica Todorović  ; Aki Vehtari  ; Patrick Rinke 􀀤   \nJ. Chem. Phys. 162, 014103 (2025)  \n[https://doi.org/10.1063/5.0229834](https://doi.org/10.1063/5.0229834)  \n􀀪  \nView Online  \n􀀮  \nExport Citation  \nArticles You May Be Interested In  \nTransfer learning for molecular property predictions from small datasets  \nAIP Advances (October 2024)  \nEntropy-based active learning of graph neural network surrogate models for materials properties  \nJ. Chem. Phys. (November 2021)  \nImplementation of automated framework in healthcare problems throughan intellectual machine learning approach  \nAIP Conf. Proc. (March 2024)  \n25 February 2025 10:45:37  \nThe Journal  \nof Chemical Physics  \nARTICLE  \n[pubs.aip.org/aip/jcp](pubs.aip.org/aip/jcp)  \nActive learning of molecular data for task-specific objectives  \n\n| Cite as: J. Chem. Phys. 162, 014103 (2025); doi: 10. 1063/5.0229834 Submitted: 19 July 2024 • Accepted: 12 December 2024 •\u003Cbr>Published Online: 2 January 2025 |  |  |  |  |\n| --- | --- | --- | --- | --- |\n| Kunal Ghosh,1 , 2  Milica Todorovi,3  Aki Vehtari,2  and Patrick Rinke1 , 4 , 5 , 6, a)  |  |  |  |  |\n| AFFILIATIONS\u003Cbr>1 Department of Applied Physics, Aalto University, P. O. Box 11000, FI-00076 Aalto, Finland\u003Cbr>2 Department of Computer Science, Aalto University, P. O. Box 15400, FI-00076 Aalto, Finland\u003Cbr>3 Department of Mechanical and Materials Engineering, University of Turku, FI-20014 Turku, Finland\u003Cbr>4 Physics Department, TUM School of Natural Sciences, Technical University of Munich, Garching, Germany\u003Cbr>5 Atomistic Modelling Center, Munich Data Science Institute, Technical University of Munich, Garching, Germany\u003Cbr>6 Munich Center for Machine Learning (MCML), Munich, Germany\u003Cbr>a)Author to whom correspondence should be addressed: [patrick.rinke@tum.de](patrick.rinke@tum.de) |  |  |  |  |\n| ABSTRACT\u003Cbr>Active learning (AL) has shown promise to be a particularly data-efficient machine learning approach. Yet, its performance depends on the application, and it is not clear when AL practitioners can expect computational savings. Here, we carry out a systematic AL performance assessment for three diverse molecular datasets and two common scientific tasks: compiling compact, informative datasets and targeted molecular searches. We implemented AL with Gaussian processes (GP) and used the many-body tensor as molecular representation. For the first task, we tested different data acquisition strategies, batch sizes, and GP noise settings. AL was insensitive to the acquisition batch size, and we observed the best AL performance for the acquisition strategy that combines uncertainty reduction with clustering to promote diversity. However, for optimal GP noise settings, AL did not outperform the randomized selection of data points. Conversely, for targeted searches, AL outperformed random sampling and achieved data savings of up to 64% . Our analysis provides insight into this task-specific performance difference in terms of target distributions and data collection strategies. We established that the performance of AL depends on the relative distribution of the target molecules in comparison to the total dataset distribution, with the largest computational savings achieved when their overlap is minimal.\u003Cbr>© 2025 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license ([https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/)). [https://doi.org/10.1063/5.0229834](https://doi.org/10.1063/5.0229834) |  |  |  |  |\n| I. INTRODUCTION\u003Cbr>In recent years, applications of machine learning (ML) in material science have produced a plethora of new discoveries,1–3 such as the discoveries of millions of novel molecules,4 phase-change materials,5 and metallic glasses.6 These discoveries rely on accurate property predictions by ML","cbCailvhvOQIFjy7","https://ap.wps.com/l/cbCailvhvOQIFjy7","pdf",6294791,1,13,"English","en",105,"# Abstract\n# Introduction\n## Motivation and background\n## Proposed approach and study scope","[{\"question\":\"What two tasks are evaluated in the active learning study?\",\"answer\":\"The study evaluates compiling compact, informative molecular datasets and conducting targeted molecular searches for specific objectives.\"},{\"question\":\"Which modeling approach is used to implement active learning?\",\"answer\":\"Active learning is implemented with Gaussian processes (GP) using the many-body tensor as the molecular representation.\"},{\"question\":\"How does active learning performance differ between compact dataset building and targeted search?\",\"answer\":\"For compact dataset building, AL may not outperform randomized selection under optimal GP noise settings. For targeted searches, AL outperforms random sampling and can achieve data savings up to 64%.\"}]","Active learning of molecular data for task-specific objectives - Research article | PDF",1785821510,33,{"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},"active-learning-of-molecular-data-for-task-specific-objectives-research-article","",{"@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/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/active-learning-of-molecular-data-for-task-specific-objectives-research-article/124309/",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},"What two tasks are evaluated in the active learning study?","Question",{"text":75,"@type":76},"The study evaluates compiling compact, informative molecular datasets and conducting targeted molecular searches for specific objectives.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which modeling approach is used to implement active learning?",{"text":80,"@type":76},"Active learning is implemented with Gaussian processes (GP) using the many-body tensor as the molecular representation.",{"name":82,"@type":73,"acceptedAnswer":83},"How does active learning performance differ between compact dataset building and targeted search?",{"text":84,"@type":76},"For compact dataset building, AL may not outperform randomized selection under optimal GP noise settings. For targeted searches, AL outperforms random sampling and can achieve data savings up to 64%.","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},[93,97,101,105,110,115,120,123,128,131,135],{"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":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},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"]