[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120516-en":3,"doc-seo-120516-105":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},120516,1374391974564,"Clementine","https://ap-avatar.wpscdn.com/avatar/14000253aa45c000a9e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779874745381141002",8,"Research & Report","Application of interpretable machine learning for cross-diagnostic inference on the ST40 spherical tokamak - Abstract - Method","Machine learning models capture complex nonlinear relationships but remain difficult to interpret and hard to guarantee safe behavior when used in high-risk settings. This limits their uptake in science and industry where interpretability matters as much as predictive accuracy. The work develops a framework that parameterises black-box models into grey-box models and applies it to plasma diagnostics using a synthetic Soft X-Ray imaging–Thomson Scattering diagnostic. The grey-box model predicts high-temporal-resolution electron temperature and density profiles from soft X-ray emission, and its outputs are benchmarked against trained black-box models across diverse plasma conditions.","arXiv :2407 . 18741v1 [physics .plasm-ph] 26 Jul 2024  \nApplication of interpretable machine learning for  \ncross-diagnostic inference on the ST40 spherical tokamak  \nT. Pyragius, C. Colgan, H. Lowe, F. Janky, M. Fontana, Y. Cai, G. Naylor, and the ST40 team  \nTokamak Energy Ltd., 173 Brook Dr, Milton, Abingdon, OX14 4SD, UK  \nE-mail: [tadas.pyragius@tokamakenergy.co.uk](tadas.pyragius@tokamakenergy.co.uk)  \nJuly 2024  \nAbstract. Machine learning models are exceptionally effective in capturing complex nonlinear relationships of high-dimensional datasets and making accurate predictions. However, their intrinsic “black-box” nature makes it difficult to interpret them or guarantee “safe behavior” when deployed in high-risk applications such as feedback control, healthcare and finance. This drawback acts as a significant barrier to their wider application across many scientific and industrial domains where the interpretability of the model predictions is as important as accuracy. Leveraging the latest developments in interpretable machine learning, we develop a method to parameterise “black-box” models, effectively transforming them into“grey-box” models. We apply this approach to plasma diagnostics by creating a parameterised synthetic Soft X-Ray imaging − Thomson Scattering diagnostic, which predicts high temporal resolution electron temperature and density profiles from the measured soft X-ray emission. The “grey-box” model predictions are benchmarked against the trained “black-box” models as well as a diverse range of plasma conditions. Our model-agnostic approach can be applied to various machine learning architectures, enabling direct comparisons of model interpretations.  \n1. Introduction  \nAdvancements in plasma physics, particularly in the context of fusion energy research, have significantly benefited from the application of machine learning techniques (see reviews [1, 2]) . These methods have proven effective in deciphering complex data generated from fusion experiments, where traditional analytic approaches can fall short. A key challenge in this domain is the interpretation of results from machine learning models, especially when these models function as “black boxes”, providing little insight into the underlying physical phenomena they model. A notable exception to this are physics informed neural networks PINNs, where the equations governing the dynamics of the system constrained by laws of physics are embedded in the loss function of the machine learning model [3] . With PINNs the model is fully interpretable because it necessarily satisfies the physics constraints. However, PINN use cases impose additional limits on their wider adoption as a new dataset input requires complete model retraining. Moreover, PINN approach is not always possible, especially if the underlying laws are unknown or are prohibitively expensive to model. In these  \n2  \nsituations a different approach is required where attempts are made to extract the underlying mechanisms learned by the model from the data.  \nA lack of model interpretability has several disadvantages. First, unlike traditional physics models that provide an analytical framework to explain interactions between physical observables and their impacts on measured outcomes, ML models offer predictions without explaining the mechanics behind them. Consequently, it becomes very difficult to discern the reasons for these predictions, thereby impeding our understanding of the governing physical mechanisms. This lack of transparency directly affects our ability to understand the underlying physics and most importantly design experiments to falsify the model and find its limitations. This issue not only erodes trust in machine learning models but also restricts their wider adoption in physics, where knowing the relationships between experimental observables is essential.  \nIn the context of plasma physics and control, the interpretability and regions of model validity are of significan","cbCailNWSzgouiWR","https://ap.wps.com/l/cbCailNWSzgouiWR","pdf",6983579,1,32,"English","en",105,"# Introduction\n## Interpretability challenges in ML\n## Impact on physics understanding and experiment design\n## Need for reliable deployment in plasma control\n## Synthetic diagnostics and limitations\n## Proposed grey-box methodology","[{\"question\":\"What problem does the document target about machine learning in high-risk scientific applications?\",\"answer\":\"It targets the black-box nature of ML models, which makes interpretations difficult and safe behavior hard to guarantee when models are deployed for mission-critical tasks like real-time plasma control.\"},{\"question\":\"How does the proposed approach turn black-box models into grey-box models?\",\"answer\":\"It parameterises explanations derived from interpretable ML techniques into a closed functional form, converting black-box behavior into a more interpretable grey-box representation without major loss of predictive performance.\"},{\"question\":\"What diagnostic task is used as a case study in this work?\",\"answer\":\"The method is applied to plasma diagnostics by building a parameterised synthetic Soft X-Ray imaging–Thomson Scattering diagnostic that predicts electron temperature and density profiles from measured soft X-ray signals.\"}]","Application of interpretable machine learning for cross-diagnostic inference on the ST40 spherical tokamak - Abstract - Method | PDF",1785730448,81,{"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},"application-of-interpretable-machine-learning-for-cross-diagnostic-inference-on-the-st40-spherical-tokamak-abstract-method","",{"@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/application-of-interpretable-machine-learning-for-cross-diagnostic-inference-on-the-st40-spherical-tokamak-abstract-method/120516/",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","2026-08-04","2026-08-03",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},"What problem does the document target about machine learning in high-risk scientific applications?","Question",{"text":76,"@type":77},"It targets the black-box nature of ML models, which makes interpretations difficult and safe behavior hard to guarantee when models are deployed for mission-critical tasks like real-time plasma control.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the proposed approach turn black-box models into grey-box models?",{"text":81,"@type":77},"It parameterises explanations derived from interpretable ML techniques into a closed functional form, converting black-box behavior into a more interpretable grey-box representation without major loss of predictive performance.",{"name":83,"@type":74,"acceptedAnswer":84},"What diagnostic task is used as a case study in this work?",{"text":85,"@type":77},"The method is applied to plasma diagnostics by building a parameterised synthetic Soft X-Ray imaging–Thomson Scattering diagnostic that predicts electron temperature and density profiles from measured soft X-ray signals.","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"]