arXiv:2511.20543v1 Announce Type: new 
Abstract: The UNet-enhanced Fourier Neural Operator (UFNO) extends the Fourier Neural Operator (FNO) by incorporating a parallel UNet pathway, enabling the retention of both high- and low-frequency components. While UFNO improves predictive accuracy over FNO, it inefficiently treats scalar inputs (e.g., temperature, injection rate) as spatially distributed fields by duplicating their values across the domain. This forces the model to process redundant constant signals within the frequency domain. Additionally, its standard loss function does not account for spatial variations in error sensitivity, limiting performance in regions of high physical importance. We introduce UFNO-FiLM, an enhanced architecture that incorporates two key innovations. First, we decouple scalar inputs from spatial features using a Feature-wise Linear Modulation (FiLM) layer, allowing the model to modulate spatial feature maps without introducing constant signals into the Fourier transform. Second, we employ a spatially weighted loss function that prioritizes learning in critical regions. Our experiments on subsurface multiphase flow demonstrate a 21\% reduction in gas saturation Mean Absolute Error (MAE) compared to UFNO, highlighting the effectiveness of our approach in improving predictive accuracy.

تم تطوير مشغل فورييه العصبي المحسن بواسطة UNet (UFNO) لتحسين التنبؤ بتدفق متعدد الأطوار في الوسائط المسامية من خلال دمج مسار UNet متوازي، مما يسمح بالاحتفاظ بمكونات التردد العالي والمنخفض. ومع ذلك، فإنه يواجه تحديات في معالجة المدخلات العددية بكفاءة، ولا تأخذ دالة الخسارة الخاصة به في الاعتبار حساسية الخطأ المكاني. تهدف UFNO-FiLM إلى معالجة هذه القيود من خلال فصل المدخلات العددية عن الميزات المكانية.

Se ha desarrollado el Operador Neural de Fourier mejorado por UNet (UFNO) para mejorar la predicción del flujo multifásico en medios porosos, incorporando una vía paralela de UNet que permite retener componentes de alta y baja frecuencia. Sin embargo, enfrenta desafíos en el procesamiento eficiente de entradas escalares y su función de pérdida no considera la sensibilidad al error espacial. La introducción de UFNO-FiLM busca abordar estas limitaciones desacoplando las entradas escalares de las características espaciales.

L'Opérateur Neural de Fourier amélioré par UNet (UFNO) a été développé pour améliorer la prédiction des flux multiphasiques dans les milieux poreux en intégrant un chemin parallèle UNet, ce qui permet de conserver à la fois les composants de haute et de basse fréquence. Cependant, il rencontre des défis dans le traitement efficace des entrées scalaires et sa fonction de perte ne prend pas en compte la sensibilité à l'erreur spatiale. L'introduction de UFNO-FiLM vise à remédier à ces limitations en découplant les entrées scalaires des caractéristiques spatiales.

The UNet-enhanced Fourier Neural Operator (UFNO) has been developed to improve the prediction of multiphase flow in porous media by incorporating a parallel UNet pathway, which retains both high- and low-frequency components. However, it faces challenges in efficiently processing scalar inputs and its loss function does not account for spatial error sensitivity. The introduction of UFNO-FiLM aims to address these limitations by decoupling scalar inputs from spatial features.

Feature-Modulated UFNO for Improved Prediction of Multiphase Flow in Porous Media

arXiv:2512.07425v1 Announce Type: cross 
Abstract: Real-time monitoring of induced seismicity is crucial for mitigating operational hazards, relying on the rapid and accurate classification of microseismic events from continuous data streams. However, while many deep learning models excel at this task, their high computational requirements often limit their practical application in real-time monitoring systems. To address this limitation, a lightweight model based on the Fourier Neural Operator (FNO) is proposed for microseismic event classification, leveraging its inherent resolution-invariance and computational efficiency for waveform processing. In the STanford EArthquake Dataset (STEAD), a global and large-scale database of seismic waveforms, the FNO-based model demonstrates high effectiveness for trigger classification, with an F1 score of 95% even in the scenario of data sparsity in training. The new FNO model greatly decreases the computer power needed relative to current deep learning models without sacrificing the classification success rate measured by the F1 score. A test on a real microseismic dataset shows a classification success rate with an F1 score of 98%, outperforming many traditional deep-learning techniques. A combination of high success rate and low computational power indicates that the FNO model can serve as a methodology of choice for real-time monitoring of microseismicity for induced seismicity. The method saves computational resources and facilitates both post-processing and real-time seismic processing suitable for the implementation of traffic light systems to prevent undesired induced seismicity.

تم اقتراح نموذج خفيف الوزن يعتمد على مشغل فورييه العصبي (FNO) لتصنيف الأحداث الزلزالية الدقيقة في الوقت الحقيقي، مما يعالج المتطلبات الحاسوبية العالية للنماذج التقليدية للتعلم العميق. وقد أظهر هذا النموذج درجة F1 ملحوظة تبلغ 95% في مجموعة بيانات زلزال ستانفورد، حتى مع ندرة بيانات التدريب.

Se ha propuesto un modelo ligero basado en el Operador Neural de Fourier (FNO) para la clasificación en tiempo real de eventos micro sísmicos, abordando las altas demandas computacionales de los modelos de aprendizaje profundo tradicionales. Este modelo ha mostrado un notable puntaje F1 del 95% en el Conjunto de Datos de Terremotos de Stanford, incluso con datos de entrenamiento escasos.

Un modèle léger basé sur l'Opérateur Neural de Fourier (FNO) a été proposé pour la classification en temps réel des événements micro-sismiques, répondant aux exigences computationnelles élevées des modèles d'apprentissage profond traditionnels. Ce modèle a montré un score F1 remarquable de 95 % dans le jeu de données STanford EArthquake, même avec des données d'entraînement rares.

A lightweight model based on the Fourier Neural Operator (FNO) has been proposed for real-time classification of microseismic events, addressing the high computational demands of traditional deep learning models. This model has shown a remarkable F1 score of 95% in the STanford EArthquake Dataset, even with sparse training data.

Feature-Modulated UFNO for Improved Prediction of Multiphase Flow in Porous Media

Was this article worth reading? Share it

PlantFCE Model Builder

Octofy

Metaflow AI