[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-119547-en":3,"doc-seo-119547-105":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},119547,1374391974564,"Clementine","https://ap-avatar.wpscdn.com/avatar/14000253aa45c000a9e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779874745381141002",8,"Research & Report","Science-Driven Atomistic Machine Learning - Review","Machine learning (ML) algorithms are becoming influential across the sciences, where the usual view frames ML as primarily data-driven. Chemistry, however, often faces the limitation of sparse, well-curated databases. This review focuses on science-driven ML approaches for atomistic modeling of materials and molecules, starting from scientific questions and selecting suitable training data and model design choices. It highlights automated, purpose-driven data collection, the use of chemical and physical priors for data efficiency, and the role of robust model evaluation and uncertainty or error estimation.","Machine Learning  \nReviews AngewandteChemie  \n[www.angewandte.org](www.angewandte.org)  \nHow to cite: Angew. Chem. Int. Ed. 2023, 62, e202219170 [doi.org/10.1002/anie.2022191](doi.org/10.1002/anie.2022191)70  \nScience-Driven Atomistic Machine Learning  \nJohannes T. Margraf*  \nAngewandte  \n Chemie  \nAngew. Chem. Int. Ed. 2023, 62, e202219170 (1 of 14) © 2023 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH  \nReviews AngewandteChemie  \n Abstract: Machine learning (ML) algorithms are currently emerging as powerful tools in all areas of science.  Conventionally, ML is understood as a fundamentally data-driven endeavour. Unfortunately, large well-curated databases are sparse in chemistry. In this contribution, I therefore review science-driven ML approaches which do not rely on “big data”, focusing on the atomistic modelling of materials and molecules. In this context, the term sciencedriven refers to approaches that begin with a scientific question and then ask what training data and model design choices are appropriate. As key features of science-driven ML, the automated and purpose-driven collection of data and the use of chemical and physical priors to achieve high data-efficiency are discussed. Furthermore, the importance of appropriate model evaluation and error estimation is emphasized.   \n1. Introduction  \nMachine learning (ML) is now an established part of several key areas of chemical research, [e.g. in](e.g. in) the development of interatomic potentials,[1, 2] the analysis of complex simulation data[3] or the design of novel drugs[4] and materials.[5] Beyond being a methodological novelty, atomistic ML has enabled real scientific breakthroughs, [e.g. in](e.g. in) predicting protein structures[6] or understanding the properties of water,[7, 8] silicon,[9] and hydrogen under extreme conditions. [10]  \nWhile chemical ML is an extraordinarily diverse subject (including applications in so-called self-driving labs[11] or in the analysis of experimental data),[12] atomistic ML is arguably one of its most mature sub-fields. Here, atomically resolved structural data serve as the main in-or outputs of a model. Among other reasons, the success of atomistic ML can be attributed to the facts that modern ML methods are inherently well suited for such high-dimensional problems, and that electronic structure calculations (most often using Density Functional Theory, DFT) offer a relatively straightforward way for generating high quality reference data.  \nIndeed, there is currently a veritable hype around ML for atomistic systems, with a multitude of new applications being reported every day. As is commonly the case with hypes, not all the reported benefits of ML hold up to scrutiny, however. For instance, comparisons with adequate (non-ML) baselines are often not performed and the applicability of the proposed methods beyond the scope of the training data is often unclear. [13]  \nHere, a certain disconnect between common practices in method development and the practical demands of atomistic modelers can be observed. For understandable reasons, the former prefer to focus on well established benchmark datasets. These are readily available and allow rigorously comparing new methods with the state-of-the-art. Unfortunately, these benchmark problems merely represent an imperfect proxy to real chemical research questions. Consequently, many proposed methods do not find their way  \n[*] Dr. J. T. Margraf  \nFritz-Haber-Institute of the Max-Planck-Society  \nFaradayweg 4–6, 14195 Berlin (Germany)  \n[E-mail: margraf@fhi.mpg.de](E-mail: margraf@fhi.mpg.de)  \n © 2023 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH. This is an open access article under the terms of the Creative Commons Attribution Non-Commercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.  \ninto practical app","cbCainORA43Te61V","https://ap.wps.com/l/cbCainORA43Te61V","pdf",3426788,1,14,"English","en",105,"# Abstract\n# Introduction\n# Big and Small Data","[{\"question\":\"What distinguishes science-driven ML from conventional data-driven ML in chemistry?\",\"answer\":\"Science-driven ML begins with a scientific question and then determines what training data and model design choices are appropriate, rather than relying mainly on large “big data” sets.\"},{\"question\":\"Why is data efficiency a key theme in science-driven atomistic ML?\",\"answer\":\"Because well-curated chemical databases are often sparse, science-driven approaches emphasize automated, purpose-driven data collection and use chemical and physical priors to reduce the data required for effective learning.\"},{\"question\":\"What evaluation aspects does the review emphasize for atomistic ML models?\",\"answer\":\"The review stresses appropriate model evaluation and emphasizes error estimation or uncertainty assessment to ensure reported performance is reliable.\"}]","Science-Driven Atomistic Machine Learning - Review | PDF",1785724891,35,{"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},"science-driven-atomistic-machine-learning-review","",{"@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/science-driven-atomistic-machine-learning-review/119547/",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","2026-08-03",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 distinguishes science-driven ML from conventional data-driven ML in chemistry?","Question",{"text":75,"@type":76},"Science-driven ML begins with a scientific question and then determines what training data and model design choices are appropriate, rather than relying mainly on large “big data” sets.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Why is data efficiency a key theme in science-driven atomistic ML?",{"text":80,"@type":76},"Because well-curated chemical databases are often sparse, science-driven approaches emphasize automated, purpose-driven data collection and use chemical and physical priors to reduce the data required for effective learning.",{"name":82,"@type":73,"acceptedAnswer":83},"What evaluation aspects does the review emphasize for atomistic ML models?",{"text":84,"@type":76},"The review stresses appropriate model evaluation and emphasizes error estimation or uncertainty assessment to ensure reported performance is reliable.","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"]