[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-128256-en":3,"doc-seo-128256-105":31,"detail-sidebar-cat-0-en-105":93},{"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":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},128256,2336475104736,"Quinn","https://ap-avatar.wpscdn.com/avatar/22000c4c5e0e5b17e70?x-image-process=image/resize,m_fixed,w_180,h_180&k=1786591360781797222",8,"Research & Report","Probabilistic Machine Learning for Real-Time Gravitational-Wave Inference - Dissertation","Gravitational-wave (GW) astronomy has delivered major discoveries in the last decade, and next-generation detectors raise expectations for further breakthroughs. Progress depends on accurately characterizing GW sources from measured data, yet standard Bayesian inference is often too computationally demanding for large-scale or real-time use. This dissertation presents DINGO, a probabilistic machine learning framework using neural posterior estimation trained on GW simulations to map data to source parameters with high speed and accuracy.","Probabilistic Machine Learning for Real-Time Gravitational-Wave Inference  \nDissertation  \nder Mathematisch-Naturwissenschaftlichen Fakultät der Eberhard Karls Universität Tübingen zur Erlangung des Grades eines Doktors der Naturwissenschaften  \n(Dr. rer. nat. )  \nvorgelegt von  \nMaximilian Dax  \naus Bonn  \nTübingen  \nGedruckt mit Genehmigung der Mathematisch-Naturwissenschaftlichen Fakultät der Eberhard Karls Universität Tübingen.  \nTag der mündlichen Qualifikation: 17.07.2025  \nDekan: Prof. Dr. Thilo Stehle  \n1. Berichterstatter/-in: Prof. Dr. Bernhard Schölkopf  \n2. Berichterstatter/-in: Prof. Dr. Philipp Hennig  \n3. Berichterstatter/-in: Prof. Dr. Tilman Plehn  \nTo You.  \nAbstract  \nGravitational-wave (GW) astronomy has led to groundbreaking discoveries in the past decade, and with the development of next-generation detectors, its potential for future breakthroughs continues to grow. This field hinges on the ability to accurately characterize GW sources based on measured data. However, computational demands of existing inference methods impede their application to large-scale or real-time data analysis. We here present DINGO, a probabilistic machine learning framework for Bayesian GW inference that addresses these limitations with an unprecedented combination of speed and accuracy. Building on neural posterior estimation (NPE), DINGO trains deep neural networks on GW simulations to learn the mapping between measured data and GW source parameters.  \nWe first introduce DINGO for binary black hole mergers, the most common GW source. We develop techniques to integrate symmetries (called GNPE) and to rapidly adapt to varying detector noise properties. We then augment NPE with importance sampling (NPE-IS) to correct for potential network inaccuracies. This enables asymptotically exact inference, independent verification and unbiased estimates of the Bayesian evidence, addressing important limitations of deep learning-based inference. Finally, we extend DINGO to binary neutron star mergers. We develop techniques to effectively compress long signals based on event-adaptive priors (prior conditioning) and to enable inference even before the merger. With inference times of less than a second, this provides crucial real-time information for directing searches for electromagnetic counterparts.  \nOur experimental evaluations encompass more than 50 real events and thousands of simulations, three different waveform models, two types of sources and two experimental setups (LIGO-Virgo-KAGRA and next-generation detectors) . DINGO consistently achieves comparable accuracy to established inference methods while being orders of magnitude faster. This prepares GW data analysis for increasing detection rates, facilitates large-scale studies and can improve searches for electromagnetic counterparts. Beyond GW astronomy, DINGO contributes several broadly applicable techniques to the field of simulation-based inference, including GNPE, NPE-IS and prior-conditioning.  \nZusammenfassung  \nDie Gravitationswellen-Astronomie hat im letzten Jahrzehnt bahnbrechende Entdeckungen ermöglicht. Mit der Entwicklung der nächsten Generation von Detektoren wächst ihr Potenzial für zukünftige Durchbrüche weiter. Ein zentraler Bestandteil dieses Forschungsfeldes ist die Charakterisierung von astrophysikalischen Gravitationswellenquellen anhand gemessener Daten. Existierende Inferenzmethoden sind allerdings so rechenintensiv, dass groß angelegte oder Echtzeitanalysen damit nur bedingt durchführbar sind. In dieser Arbeit präsentieren wir DINGO, ein probabilistisches System des maschinellen Lernens für Bayessche Inferenz von Gravitationswellen, welches diese Einschränkungen überwindet. Aufbauend auf der Methode der neuronalen Posteriorschätzung (engl. neural posterior estimation, NPE) trainiert DINGO tiefe neuronale Netzwerke mit Simulationen von Gravitationswellen, und lernt so die Zusammenhänge zwischen gemessenen Daten und Parametern, welche die Gravitationswellenquellen","cbCaiutdzAIhSx7E","https://ap.wps.com/l/cbCaiutdzAIhSx7E","pdf",31289223,6,1,160,"English","en",105,"# Abstract\n## DINGO framework and neural posterior estimation\n## Techniques: GNPE and noise adaptation\n## Importance sampling correction (NPE-IS)\n## Extension to binary neutron star mergers","[{\"question\":\"What problem does DINGO address in gravitational-wave inference?\",\"answer\":\"Existing inference methods are computationally expensive, limiting large-scale or real-time analysis. DINGO targets this bottleneck by combining probabilistic machine learning with fast neural inference.\"},{\"question\":\"How does DINGO learn the relationship between data and source parameters?\",\"answer\":\"DINGO builds on neural posterior estimation (NPE) by training deep neural networks on gravitational-wave simulations, learning the mapping from measured data to source parameters.\"},{\"question\":\"What extensions and correction techniques make DINGO more reliable?\",\"answer\":\"For binary black hole mergers it incorporates symmetries via GNPE and adapts to detector noise; it also adds importance sampling (NPE-IS) to correct for possible network inaccuracies and enable asymptotically exact inference.\"}]","Probabilistic Machine Learning for Real-Time Gravitational-Wave Inference - Dissertation | PDF",1785946275,403,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":88,"head_meta":90,"extra_data":92,"updated_unix":29},"probabilistic-machine-learning-for-real-time-gravitational-wave-inference-dissertation","",{"@graph":37,"@context":87},[38,55,70],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,49,52],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":48},"https://docshare.wps.com/document/","Document",2,{"item":50,"name":12,"@type":44,"position":51},"https://docshare.wps.com/document/research-report/",3,{"item":53,"name":13,"@type":44,"position":54},"https://docshare.wps.com/document/probabilistic-machine-learning-for-real-time-gravitational-wave-inference-dissertation/128256/",4,{"url":53,"name":13,"@type":56,"author":57,"headline":13,"publisher":59,"fileFormat":62,"inLanguage":24,"description":14,"dateModified":63,"datePublished":64,"encodingFormat":62,"isAccessibleForFree":65,"interactionStatistic":66},"DigitalDocument",{"name":9,"@type":58},"Person",{"url":42,"name":60,"@type":61},"DocShare","Organization","application/pdf","2026-08-28","2026-08-05",true,{"@type":67,"interactionType":68,"userInteractionCount":20},"InteractionCounter",{"@type":69},"ViewAction",{"@type":71,"mainEntity":72},"FAQPage",[73,79,83],{"name":74,"@type":75,"acceptedAnswer":76},"What problem does DINGO address in gravitational-wave inference?","Question",{"text":77,"@type":78},"Existing inference methods are computationally expensive, limiting large-scale or real-time analysis. DINGO targets this bottleneck by combining probabilistic machine learning with fast neural inference.","Answer",{"name":80,"@type":75,"acceptedAnswer":81},"How does DINGO learn the relationship between data and source parameters?",{"text":82,"@type":78},"DINGO builds on neural posterior estimation (NPE) by training deep neural networks on gravitational-wave simulations, learning the mapping from measured data to source parameters.",{"name":84,"@type":75,"acceptedAnswer":85},"What extensions and correction techniques make DINGO more reliable?",{"text":86,"@type":78},"For binary black hole mergers it incorporates symmetries via GNPE and adapts to detector noise; it also adds importance sampling (NPE-IS) to correct for possible network inaccuracies and enable asymptotically exact inference.","https://schema.org",{"og:url":53,"og:type":89,"og:title":13,"og:site_name":60,"og:description":14},"article",{"robots":91,"canonical":53},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":94},[95,99,103,107,112,116,121,124,129,132,136],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":96,"show_sort_weight":97,"slug":98},"Story & Novel",90,"story-novel",{"id":48,"doc_module":4,"doc_module_name":47,"category_name":100,"show_sort_weight":101,"slug":102},"Literature",80,"literature",{"id":54,"doc_module":4,"doc_module_name":47,"category_name":104,"show_sort_weight":105,"slug":106},"Exam",70,"exam",{"id":108,"doc_module":4,"doc_module_name":47,"category_name":109,"show_sort_weight":110,"slug":111},5,"Comic",60,"comic",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":113,"show_sort_weight":114,"slug":115},"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":47,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":47,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":47,"category_name":138,"show_sort_weight":108,"slug":139},19,"General","general"]