[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-116842-en":3,"doc-seo-116842-105":30,"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":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},116842,1099513958607,"Jiven","https://ap-avatar.wpscdn.com/avatar/100002390cf8733938c?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778829742770036399",8,"Research & Report","Representations of Materials for Machine Learning - Review of Context-Dependent Representations","High-throughput data generation and machine learning accelerate computational materials discovery by learning relationships among composition, structure, and properties and applying them to design. Building these connections requires translating materials data into numerical representations suitable for ML models. The survey covers representation construction across varied data formats, dataset sizes, and fidelity levels, and compares predictive modeling scopes and target properties. It also discusses representation learning, transferable chemical and physical knowledge between tasks, and high-impact open questions needing further investigation.","arXiv :2301 .08813v1 [ cond-mat .mtrl-sci ] 20 Jan 2023  \nRepresentations of Materials for Machine Learning ∗  \nJames Damewood 1 , Jessica Karaguesian 1,2 , Jaclyn R. Lunger 1 , Aik Rui Tan 1 , Mingrou Xie 1,3 ,  \nJiayu Peng 1 , and Rafael G􀀓omez-Bombarelli†1  \n1 Department of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, USA, 02129  \n2 Center for Computational Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA, USA, 02139  \n3 Department of Chemical Engineering, Massachusetts Institute of Technology, 77  \nMassachusetts Avenue, Cambridge, MA, USA, 02139  \nJanuary 24, 2023  \nAbstract  \nHigh-throughput data generation methods and machine learning (ML) algorithms have given rise to a new era of computational materials science by learning relationships among composition, structure, and properties and by exploiting such relations for design. However, to build these connections, materials data must be translated into a numerical form, called a representation, that can be processed by a machine learning model. Datasets in materials science vary in format (ranging from images to spectra), size, and 􀀌delity. Predictive models vary in scope and property of interests. Here, we review context-dependent strategies for constructing representations that enable the use of materials as inputs or outputs of machine learning models. Furthermore, we discuss how modern ML techniques can learn representations from data and transfer chemical and physical information between tasks. Finally, we outline high-impact questions that have not been fully resolved and thus, require further investigation.  \nContents  \n1 INTRODUCTION 2  \n2 STRUCTURAL FEATURES FOR ATOMISTIC GEOMETRIES 3  \n2.1 Local Descriptors ........................................ 3  \n2.2 Global Descriptors ....................................... 5  \n2.3 Topological Descriptors .................................... 7  \n3 LEARNING ON PERIODIC CRYSTAL GRAPHS 8  \n4 CONSTRUCTING REPRESENTATIONS FROM STOICHIOMETRY 11  \n5 DEFECTS, SURFACES, AND GRAIN BOUNDARIES 13  \n6 TRANSFERABLE INFORMATION BETWEEN REPRESENTATIONS 15  \n∗ Accepted for publication in Annual Review of Materials Research Volume 53, [https://www.annualreviews.org/](https://www.annualreviews.org/) .†[rafagb@mit.edu](rafagb@mit.edu)  \n7 GENERATIVE MODELS FOR INVERSE DESIGN 17  \n8 DISCUSSION 19  \n8.1 Trade-o􀀋s of Local and Global Structural Descriptors ................... 19  \n8.2 Prediction from Unrelaxed Crystal Prototypes ........................ 19  \n8.3 Applicability of Compositional Descriptors .......................... 20  \n8.4 Extensions of Generative Models ............................... 20  \n1 INTRODUCTION  \nEnergy and sustainability applications demand the rapid development of scalable new materials technologies. Big data and machine learning (ML) have been proposed as strategies to rapidly identify \\needle-in-the-haystack\" materials that have the potential for revolutionary impact.  \nHigh-throughput experimentation platforms based on robotized laboratories can increase the e􀀎 -ciency and speed of synthesis and characterization. However, in many practical open problems, the number of possible design parameters is too large to be analyzed exhaustively. Virtual screening somewhat mitigates this challenge by using physics-based simulations to suggest the most promising candidates, reducing the cost but also the 􀀌delity of the screens[1, 2, 3] .  \nOver the past decade, hardware improvements, new algorithms, and the development of large-scale repositories of materials data [4, 5, 6, 7, 8, 9] have enabled a new era of ML methods. In principle, predictive ML models can identify and exploit nontrivial trends in high-dimensional data to achieve accuracy comparable with or superior to 􀀌rst-principles calculations, but with orders of magnitude reduction in cost. In practice, while a judicious model choice is helpful in mov","cbCaioIC1DwpAp9d","https://ap.wps.com/l/cbCaioIC1DwpAp9d","pdf",10513440,1,31,"English","en",105,"# Introduction\n# Structural features for atomistic geometries\n## Local descriptors\n## Global descriptors\n## Topological descriptors\n# Learning on periodic crystal graphs\n# Constructing representations from stoichiometry\n# Defects, surfaces, and grain boundaries\n# Transferable information between representations\n# Generative models for inverse design\n# Discussion\n## Trade-offs of local and global structural descriptors\n## Prediction from unrelaxed crystal prototypes\n## Applicability of compositional descriptors\n## Extensions of generative models","[{\"question\":\"Why are numerical representations necessary for machine learning in materials science?\",\"answer\":\"Materials data must be translated into numerical representations so ML models can process composition, structure, and properties. Representations must include features and descriptors from which relevant physics and chemistry can emerge.\"},{\"question\":\"How do representation strategies account for different dataset formats and fidelity levels?\",\"answer\":\"Materials datasets vary in format (e.g., images versus spectra), size, and fidelity, so representation construction must be context-dependent. The survey reviews strategies that enable these diverse data sources to serve as inputs or outputs for ML models.\"},{\"question\":\"How can modern ML learn representations and transfer information between tasks?\",\"answer\":\"The document discusses learning representations from data and transferring chemical and physical information across tasks. This supports more effective use of learned structure-property relationships in different predictive settings.\"}]","Representations of Materials for Machine Learning - Review of Context-Dependent Representations | PDF",1785672035,78,{"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":87,"head_meta":89,"extra_data":91,"updated_unix":28},"representations-of-materials-for-machine-learning-review-of-context-dependent-representations","",{"@graph":36,"@context":86},[37,54,69],{"@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/representations-of-materials-for-machine-learning-review-of-context-dependent-representations/116842/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05","2026-08-02",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 are numerical representations necessary for machine learning in materials science?","Question",{"text":76,"@type":77},"Materials data must be translated into numerical representations so ML models can process composition, structure, and properties. Representations must include features and descriptors from which relevant physics and chemistry can emerge.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How do representation strategies account for different dataset formats and fidelity levels?",{"text":81,"@type":77},"Materials datasets vary in format (e.g., images versus spectra), size, and fidelity, so representation construction must be context-dependent. The survey reviews strategies that enable these diverse data sources to serve as inputs or outputs for ML models.",{"name":83,"@type":74,"acceptedAnswer":84},"How can modern ML learn representations and transfer information between tasks?",{"text":85,"@type":77},"The document discusses learning representations from data and transferring chemical and physical information across tasks. This supports more effective use of learned structure-property relationships in different predictive settings.","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":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]