arXiv:2511.09118v2 Announce Type: replace 
Abstract: Generative models are increasingly central to scientific workflows, yet their systematic use and interpretation require a proper understanding of their limitations through rigorous validation. Classic approaches struggle with scalability, statistical power, or interpretability when applied to high-dimensional data, making it difficult to certify the reliability of these models in realistic, high-dimensional scientific settings. Here, we propose the use of the New Physics Learning Machine (NPLM), a learning-based approach to goodness-of-fit testing inspired by the Neyman--Pearson construction, to test generative networks trained on high-dimensional scientific data. We demonstrate the performance of NPLM for validation in two benchmark cases: generative models trained on mixtures of Gaussian models with increasing dimensionality, and a public end-to-end model, known as FlowSim, developed to generate high-energy physics collision events. We demonstrate that the NPLM can serve as a powerful validation method while also providing a means to diagnose sub-optimally modeled regions of the data.

تم اقتراح نهج جديد للتحقق من صحة النماذج التوليدية، يُعرف باسم آلة التعلم الفيزيائية الجديدة (NPLM)، لمعالجة قيود الأساليب التقليدية في البيانات العلمية عالية الأبعاد. يستلهم هذا الأسلوب من بناء نيمان-بيرسون ويهدف إلى تعزيز موثوقية الشبكات التوليدية، خاصة في سير العمل العلمي المعقد.

Se ha propuesto un nuevo enfoque para validar modelos generativos, denominado New Physics Learning Machine (NPLM), para abordar las limitaciones de los métodos tradicionales en datos científicos de alta dimensión. Este método, inspirado en la construcción de Neyman-Pearson, tiene como objetivo mejorar la fiabilidad de las redes generativas, especialmente en flujos de trabajo científicos complejos.

Une nouvelle approche pour valider les modèles génératifs, appelée New Physics Learning Machine (NPLM), a été proposée pour surmonter les limitations des méthodes traditionnelles dans les données scientifiques de haute dimension. Cette méthode, inspirée de la construction de Neyman-Pearson, vise à améliorer la fiabilité des réseaux génératifs, en particulier dans des flux de travail scientifiques complexes.

A new approach to validating generative models, called the New Physics Learning Machine (NPLM), has been proposed to address the limitations of traditional methods in high-dimensional scientific data. This method is inspired by the Neyman-Pearson construction and aims to enhance the reliability of generative networks, particularly in complex scientific workflows.

Learning to Validate Generative Models: a Goodness-of-Fit Approach

One More Thing in AI – Your Shortcut to AI Mastery

Learning to Validate Generative Models: a Goodness-of-Fit Approach

Was this article worth reading? Share it

One More Thing in AI

LucidQuery AI

Magicley AI

Dynamiq

AIPortalX

FastML

Ready to build your own newsroom?