[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123978-en":3,"doc-seo-123978-105":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":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},123978,8796095462418,"Noah","https://ap-avatar.wpscdn.com/avatar/80000253c1241d02b47?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778826106357471780",8,"Research & Report","Beyond development - Challenges in deploying machine learning models for structural engineering applications","Machine learning (ML) solutions are transforming structural engineering, yet most methods remain proof-of-concept and rarely reach real-world deployment. This paper explains development-to-deployment challenges using two illustrative examples, highlighting common pitfalls such as model overfitting and underspecification, inadequate training data representativeness, variable omission bias, and reliance on cross-validation without sufficient safeguards. The discussion emphasizes rigorous validation and deployment-oriented design via adaptive sampling, careful physics-informed feature selection, and balancing model complexity with generalizability.","arXiv :2404 . 12544v1 [ cs .LG] 18 Apr 2024  \nBeyond development: Challenges in deploying machine learning models for  \nstructural engineering applications  \nMohsen Zaker Esteghamati 1 , Brennan Bean2 , Henry V. Burton3 , and M.Z. Naser4  \n1Department of Civil and Environmental Engineering, Utah State University, Logan, UT. Email: [mohsen.zaker@usu.edu](mohsen.zaker@usu.edu)  \n2Department of Mathematics and Statistics, Utah State University, Logan, UT.  \n3Department of Civil and Environmental Engineering, University of California at Los Angeles, Los Angeles, CA.  \n4 School of Civil and Environmental Engineering and Earth Sciences, Clemson University, Clemson, SC.  \nABSTRACT  \nMachine learning (ML)-based solutions are rapidly changing the landscape of many fields, including structural engineering. Despite their promising performance, these approaches are usually only demonstrated as proof-of-concept in structural engineering, and are rarely deployed for real-world applications. This paper aims to illustrate the challenges of developing ML models suitable for deployment through two illustrative examples. Among various pitfalls, the presented discussion focuses on model overfitting and underspecification, training data representativeness, variable omission bias, and cross-validation. The results highlight the importance of implementing rigorous model validation techniques through adaptive sampling, careful physics-informed featureselection, and considerations of both model complexity and generalizability.  \nINTRODUCTION  \nThe last decade witnessed a surge in machine learning (ML)-based solutions to address various structural engineering problems, ranging from analyzing (Qin and Naser 2023; Xu et al. 2022 ;  \n1 Zaker Esteghamati et al, April 22, 2024  \nHwang et al. 2021 ; Kourehpaz and Molina Hutt 2022) to designing (Zhao et al. 2023 ; Esteghamati and Flint 2021 ; Esteghamati and Flint 2023; Issa et al. 2023) and monitoring (Mohammadi et al. 2023 ; Soleimani-Babakamali et al. 2023 ; Yuan et al. 2023 ; Bashar and Torres-Machi 2022) the built environment. The current practice of structural engineering-oriented model development is heavily focused on identifying the most “accurate” ML model. This emphasis is mainly due to the justification of ML as an alternative to complex and time-consuming methods, where ML provides a more accurate solution at lower computational cost and time.  \nThe life cycle of an ML model has two main stages of development and deployment (Figure 1) . Most research papers rightfully focus on model development, though the practical benefits of the ML model are achieved during deployment. However, despite the successful implementations of several ML-based solutions for various structural engineering problems, little attention has been given to exploring the efficacy of these models beyond \"establishing a proof-of-concept\". Such explorations are critical as the successful development of ML models does not necessarily translate into useful solutions that can be deployed for real-world datasets (Baier et al. 2019) .  \nFundamentally, almost all ML models (due to their statistical nature, and similar to other wellestablished approaches such as empirical analysis) capture data association rather than causal relationships. Here, data association refers to one variable providing information about another variable. In contrast, a causal relationship occurs when one variable results from another variable, and its absence leads to a counterfactual statement (Naser and Çiftçioğlu 2023; Naser 2022; Burton 2023 ; Burton and Baker 2023) . This means that the variables identified by an ML model as“important” for accurate predictions do not necessarily are variables that drive the engineering process in question. Even when ignoring the limitations of ML models in identifying causal relationships, many ML models make it difficult to determine the nature of the association between variables. Such a \"black-box\" aspect negatively aff","cbCaifqBoK43n4Lq","https://ap.wps.com/l/cbCaifqBoK43n4Lq","pdf",2500137,1,28,"English","en",105,"# Abstract\n# Introduction\n## Model lifecycle: development vs deployment\n## Data association vs causal relationships\n## Limitations of accuracy metrics in deployment\n## Over-sampling and generalization beyond training\n## Mapping ML engineering themes to structural engineering challenges","[{\"question\":\"为什么结构工程中的ML方案往往难以从研究走向真实部署？\",\"answer\":\"文中指出，很多方法只完成“证明概念”，而未充分考虑开发结果在真实数据上的可靠性与可泛化性。常见问题包括过拟合/欠说明、训练数据不具代表性以及变量遗漏偏差。\"},{\"question\":\"文章如何解释ML模型的“黑箱”与用户信任之间的关系？\",\"answer\":\"文章强调，ML主要捕捉数据关联而非因果关系，且难以判断变量关联的本质。该“黑箱”特性会削弱用户在部署时的信心。\"},{\"question\":\"为提升部署可靠性，文中提出了哪些验证与建模要点？\",\"answer\":\"文中强调通过自适应采样进行更严格的模型验证，并结合基于物理的特征选择，同时权衡模型复杂度与泛化能力，从而减少开发与部署之间的脱节。\"}]","Beyond development - Challenges in deploying machine learning models for structural engineering applications | PDF",1785819555,71,{"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},"beyond-development-challenges-in-deploying-machine-learning-models-for-structural-engineering-applications","",{"@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/beyond-development-challenges-in-deploying-machine-learning-models-for-structural-engineering-applications/123978/",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":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"为什么结构工程中的ML方案往往难以从研究走向真实部署？","Question",{"text":75,"@type":76},"文中指出，很多方法只完成“证明概念”，而未充分考虑开发结果在真实数据上的可靠性与可泛化性。常见问题包括过拟合/欠说明、训练数据不具代表性以及变量遗漏偏差。","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"文章如何解释ML模型的“黑箱”与用户信任之间的关系？",{"text":80,"@type":76},"文章强调，ML主要捕捉数据关联而非因果关系，且难以判断变量关联的本质。该“黑箱”特性会削弱用户在部署时的信心。",{"name":82,"@type":73,"acceptedAnswer":83},"为提升部署可靠性，文中提出了哪些验证与建模要点？",{"text":84,"@type":76},"文中强调通过自适应采样进行更严格的模型验证，并结合基于物理的特征选择，同时权衡模型复杂度与泛化能力，从而减少开发与部署之间的脱节。","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"]