arXiv:2510.23396v1 Announce Type: new 
Abstract: The immense success of the Transformer architecture
  in Natural Language Processing has led to its adoption in Time Se ries Forecasting (TSF), where superior performance has been shown.
  However, a recent important paper questioned their effectiveness by
  demonstrating that a simple single layer linear model outperforms
  Transformer-based models. This was soon shown to be not as valid,
  by a better transformer-based model termed PatchTST. More re cently, TimeLLM demonstrated even better results by repurposing a
  Large Language Model (LLM) for the TSF domain. Again, a follow
  up paper challenged this by demonstrating that removing the LLM
  component or replacing it with a basic attention layer in fact yields
  better performance. One of the challenges in forecasting is the fact
  that TSF data favors the more recent past, and is sometimes subject
  to unpredictable events. Based upon these recent insights in TSF, we
  propose a strong Mixture of Experts (MoE) framework. Our method
  combines the state-of-the-art (SOTA) models including xLSTM, en hanced Linear, PatchTST, and minGRU, among others. This set of
  complimentary and diverse models for TSF are integrated in a Trans former based MoE gating network. Our proposed model outperforms
  all existing TSF models on standard benchmarks, surpassing even the
  latest approaches based on MoE frameworks.

ناقشت مناقشة حديثة في مجال التنبؤ بسلاسل الزمن (TSF) الجدل المستمر حول فعالية نماذج Transformer مقارنة بالنماذج الخطية البسيطة. بينما ادعى مقال سابق أن نموذجًا خطيًا أساسيًا يتفوق على Transformers، تم تحدي هذا الادعاء من خلال تقديم نموذج جديد يسمى PatchTST، الذي أظهر أداءً متفوقًا. هذه المحادثة مهمة لأنها تعكس المشهد المتطور لتقنيات التنبؤ وأهمية الابتكار المستمر في تطوير النماذج.

Una reciente discusión en el campo de la previsión de series temporales (TSF) destacó el debate en curso sobre la efectividad de los modelos Transformer en comparación con modelos lineales más simples. Mientras que un artículo anterior afirmaba que un modelo lineal básico superaba a los Transformers, esta afirmación fue desafiada por la introducción de un nuevo modelo llamado PatchTST, que demostró un rendimiento superior. Esta conversación es significativa ya que refleja el paisaje en evolución de las técnicas de previsión y la importancia de la innovación continua en el desarrollo de modelos.

Une récente discussion dans le domaine de la prévision de séries temporelles (TSF) a mis en lumière le débat en cours sur l'efficacité des modèles Transformer par rapport aux modèles linéaires plus simples. Bien qu'un article précédent ait affirmé qu'un modèle linéaire de base surpassait les Transformers, cette assertion a été contestée par l'introduction d'un nouveau modèle appelé PatchTST, qui a démontré des performances supérieures. Cette conversation est importante car elle reflète l'évolution des techniques de prévision et l'importance de l'innovation continue dans le développement des modèles.

A recent discussion in the field of Time Series Forecasting (TSF) highlighted the ongoing debate about the effectiveness of Transformer models compared to simpler linear models. While an earlier paper claimed that a basic linear model outperformed Transformers, this assertion was challenged by the introduction of a new model called PatchTST, which demonstrated superior performance. This conversation is significant as it reflects the evolving landscape of forecasting techniques and the importance of continuous innovation in model development.

EMTSF:Extraordinary Mixture of SOTA Models for Time Series Forecasting

arXiv:2512.07590v1 Announce Type: new 
Abstract: To address the challenge of segmenting noisy images with blurred or fragmented boundaries, this paper presents a robust version of Variational Model Based Tailored UNet (VM_TUNet), a hybrid framework that integrates variational methods with deep learning. The proposed approach incorporates physical priors, an edge detector and a mean curvature term, into a modified Cahn-Hilliard equation, aiming to combine the interpretability and boundary-smoothing advantages of variational partial differential equations (PDEs) with the strong representational ability of deep neural networks. The architecture consists of two collaborative modules: an F module, which conducts efficient frequency domain preprocessing to alleviate poor local minima, and a T module, which ensures accurate and stable local computations, backed by a stability estimate. Extensive experiments on three benchmark datasets indicate that the proposed method achieves a balanced trade-off between performance and computational efficiency, which yields competitive quantitative results and improved visual quality compared to pure convolutional neural network (CNN) based models, while achieving performance close to that of transformer-based method with reasonable computational expense.

تقدم دراسة جديدة نسخة قوية من نموذج UNet المخصص القائم على النماذج التغيرية (VM_TUNet)، الذي يدمج الأساليب التغيرية مع التعلم العميق لتحسين تقسيم الصور، خاصة في الصور المزعجة ذات الحدود الضبابية. يستخدم الإطار كاشف حواف وعبارة انحناء متوسط ضمن معادلة كاهن-هيلارد المعدلة، مما يظهر أداءً محسنًا من خلال وحدتين تعاونية لمعالجة مسبقة فعالة وحسابات محلية مستقرة.

Un nuevo estudio presenta una versión robusta del Variational Model Based Tailored UNet (VM_TUNet), que integra métodos variacionales con aprendizaje profundo para mejorar la segmentación de imágenes, especialmente en imágenes ruidosas con bordes difusos. El marco emplea un detector de bordes y un término de curvatura media dentro de una ecuación de Cahn-Hilliard modificada, demostrando un rendimiento mejorado a través de dos módulos colaborativos para un preprocesamiento eficiente y cálculos locales estables.

Une nouvelle étude présente une version robuste du Variational Model Based Tailored UNet (VM_TUNet), qui intègre des méthodes variationnelles avec l'apprentissage profond pour améliorer la segmentation d'images, en particulier dans les images bruyantes avec des contours flous. Le cadre utilise un détecteur de contours et un terme de courbure moyenne dans une équation de Cahn-Hilliard modifiée, démontrant une performance améliorée grâce à deux modules collaboratifs pour un prétraitement efficace et des calculs locaux stables.

A new study introduces a robust version of the Variational Model Based Tailored UNet (VM_TUNet), which integrates variational methods with deep learning to enhance image segmentation, particularly in noisy images with blurred boundaries. The framework employs an edge detector and a mean curvature term within a modified Cahn-Hilliard equation, demonstrating improved performance through two collaborative modules for efficient preprocessing and stable local computations.

Robust Variational Model Based Tailored UNet: Leveraging Edge Detector and Mean Curvature for Improved Image Segmentation

arXiv:2512.07747v1 Announce Type: new 
Abstract: Unified understanding and generation is a highly appealing research direction in multimodal learning. There exist two approaches: one trains a transformer via an auto-regressive paradigm, and the other adopts a two-stage scheme connecting pre-trained understanding and generative models for alignment fine-tuning. The former demands massive data and computing resources unaffordable for ordinary researchers. Though the latter requires a lower training cost, existing works often suffer from limited task coverage or poor generation quality. Both approaches lack the ability to parse input meta-information (such as task type, image resolution, video duration, etc.) and require manual parameter configuration that is tedious and non-intelligent. In this paper, we propose Unison which adopts the two-stage scheme while preserving the capabilities of the pre-trained models well. With an extremely low training cost, we cover a variety of multimodal understanding tasks, including text, image, and video understanding, as well as diverse generation tasks, such as text-to-visual content generation, editing, controllable generation, and IP-based reference generation. We also equip our model with the ability to automatically parse user intentions, determine the target task type, and accurately extract the meta-information required for the corresponding task. This enables full automation of various multimodal tasks without human intervention. Experiments demonstrate that, under a low-cost setting of only 500k training samples and 50 GPU hours, our model can accurately and automatically identify tasks and extract relevant parameters, and achieve superior performance across a variety of understanding and generation tasks.

تم تقديم إطار عمل جديد يسمى Unison، مصمم لفهم وتوليد موحد في التعلم متعدد الوسائط. يعتمد هذا الإطار على مخطط من مرحلتين يستخدم بفعالية نماذج مدربة مسبقًا مع تقليل تكاليف التدريب بشكل كبير، مما يعالج قيود الأساليب الحالية التي تتطلب إما بيانات ضخمة أو تعاني من ضعف جودة التوليد.

Se ha presentado un nuevo marco llamado Unison, diseñado para la comprensión y generación unificadas en el aprendizaje multimodal. Este marco adopta un esquema de dos etapas que utiliza eficazmente modelos preentrenados mientras reduce significativamente los costos de entrenamiento, abordando las limitaciones de los enfoques existentes que requieren grandes cantidades de datos o sufren de mala calidad en la generación.

Un nouveau cadre nommé Unison a été introduit, conçu pour la compréhension et la génération unifiées dans l'apprentissage multimodal. Ce cadre adopte un schéma en deux étapes qui utilise efficacement des modèles pré-entraînés tout en réduisant considérablement les coûts d'entraînement, répondant ainsi aux limitations des approches existantes qui nécessitent soit des données extensives, soit souffrent d'une mauvaise qualité de génération.

A new framework named Unison has been introduced, designed for unified understanding and generation in multimodal learning. This framework adopts a two-stage scheme that effectively utilizes pre-trained models while significantly reducing training costs, addressing the limitations of existing approaches that either require extensive data or suffer from poor generation quality.

EMTSF:Extraordinary Mixture of SOTA Models for Time Series Forecasting

Was this article worth reading? Share it

Airparser

LucidQuery AI

Hypertune