arXiv:2511.16986v1 Announce Type: new 
Abstract: Radiomap serves as a vital tool for wireless network management and deployment by providing powerful spatial knowledge of signal propagation and coverage. However, increasingly complex radio propagation behavior and surrounding environments pose strong challenges for radiomap estimation (RME). In this work, we propose a knowledge-guided RME framework that integrates Kolmogorov-Arnold Networks (KAN) with Mixture-of-Experts (MoE), namely RadioKMoE. Specifically, we design a KAN module to predict an initial coarse coverage map, leveraging KAN's strength in approximating physics models and global radio propagation patterns. The initial coarse map, together with environmental information, drives our MoE network for precise radiomap estimation. Unlike conventional deep learning models, the MoE module comprises expert networks specializing in distinct radiomap patterns to improve local details while preserving global consistency. Experimental results in both multi- and single-band RME demonstrate the enhanced accuracy and robustness of the proposed RadioKMoE in radiomap estimation.

التطور الأخير لـ RadioKMoE يقدم إطارًا موجهًا بالمعرفة لتقدير خرائط الإشعاع، حيث يجمع بين شبكات كولموغوروف-أرنولد (KAN) ومزيج من الخبراء (MoE) لتحسين إدارة الشبكات اللاسلكية. يهدف هذا النهج المبتكر إلى معالجة التحديات التي تطرحها سلوكيات انتشار الإشعاع المعقدة والبيئات المحيطة، مع السعي للحصول على توقعات أكثر دقة لتغطية الإشارة.

El reciente desarrollo de RadioKMoE introduce un marco guiado por el conocimiento para la estimación de radiomapas, combinando Redes de Kolmogorov-Arnold (KAN) con Mixtures of Experts (MoE) para mejorar la gestión de redes inalámbricas. Este enfoque innovador aborda los desafíos planteados por los comportamientos complejos de propagación de radio y los entornos, buscando predicciones más precisas de cobertura de señal.

Le développement récent de RadioKMoE introduit un cadre guidé par les connaissances pour l'estimation des radiomaps, combinant les réseaux de Kolmogorov-Arnold (KAN) avec des Mixtures d'Experts (MoE) pour améliorer la gestion des réseaux sans fil. Cette approche innovante vise à relever les défis posés par les comportements complexes de propagation radio et les environnements, en visant des prédictions de couverture de signal plus précises.

The recent development of RadioKMoE introduces a knowledge-guided framework for radiomap estimation, combining Kolmogorov-Arnold Networks (KAN) with Mixture-of-Experts (MoE) to enhance wireless network management. This innovative approach addresses the challenges posed by complex radio propagation behaviors and environments, aiming for more accurate signal coverage predictions.

RadioKMoE: Knowledge-Guided Radiomap Estimation with Kolmogorov-Arnold Networks and Mixture-of-Experts

arXiv:2511.11236v2 Announce Type: replace 
Abstract: In recent years, image editing has garnered growing attention. However, general image editing models often fail to produce satisfactory results when confronted with new styles. The challenge lies in how to effectively fine-tune general image editing models to new styles using only a limited amount of paired data. To address this issue, this paper proposes a novel few-shot style editing framework. For this task, we construct a benchmark dataset that encompasses five distinct styles. Correspondingly, we propose a parameter-efficient multi-style Mixture-of-Experts Low-Rank Adaptation (MoE LoRA) with style-specific and style-shared routing mechanisms for jointly fine-tuning multiple styles. The style-specific routing ensures that different styles do not interfere with one another, while the style-shared routing adaptively allocates shared MoE LoRAs to learn common patterns. Our MoE LoRA can automatically determine the optimal ranks for each layer through a novel metric-guided approach that estimates the importance score of each single-rank component. Additionally, we explore the optimal location to insert LoRA within the Diffusion in Transformer (DiT) model and integrate adversarial learning and flow matching to guide the diffusion training process. Experimental results demonstrate that our proposed method outperforms existing state-of-the-art approaches with significantly fewer LoRA parameters.

تم اقتراح إطار جديد لتحرير الأنماط القليل من اللقطات لتحسين قدرات تحرير الصور، حيث يتناول القيود المفروضة على النماذج العامة في التكيف مع الأنماط الجديدة باستخدام بيانات مقترنة محدودة. يستخدم هذا الإطار تكيف منخفض الرتبة لمزيج من الخبراء متعدد الأنماط (MoE LoRA) الذي يتضمن آليات توجيه محددة للأنماط ومشتركة لتعديل فعال عبر أنماط متعددة.

Se ha propuesto un nuevo marco de edición de estilo de pocos disparos para mejorar las capacidades de edición de imágenes, abordando las limitaciones de los modelos generales para adaptarse a nuevos estilos con datos emparejados mínimos. Este marco utiliza una adaptación de bajo rango de mezcla de expertos multiestilo (MoE LoRA) que incorpora mecanismos de enrutamiento específicos y compartidos por estilo para un ajuste efectivo en múltiples estilos.

Un nouveau cadre d'édition de style peu coûteux a été proposé pour améliorer les capacités d'édition d'images, abordant les limitations des modèles généraux à s'adapter à de nouveaux styles avec des données appariées minimales. Ce cadre utilise une adaptation à faible rang de mélange d'experts multi-styles (MoE LoRA) efficace en paramètres, qui intègre à la fois des mécanismes de routage spécifiques au style et partagés pour un ajustement efficace à travers plusieurs styles.

A novel few-shot style editing framework has been proposed to enhance image editing capabilities, addressing the limitations of general models in adapting to new styles with minimal paired data. This framework utilizes a parameter-efficient multi-style Mixture-of-Experts Low-Rank Adaptation (MoE LoRA) that incorporates both style-specific and style-shared routing mechanisms for effective fine-tuning across multiple styles.

Parameter-Efficient MoE LoRA for Few-Shot Multi-Style Editing

arXiv:2509.23366v2 Announce Type: replace 
Abstract: Feature selection is a key step in many tabular prediction problems, where multiple candidate variables may be redundant, noisy, or weakly informative. We investigate feature selection based on Kolmogorov-Arnold Networks (KANs), which parameterize feature transformations with splines and expose per-feature importance scores in a natural way. From this idea we derive four KAN-based selection criteria (coefficient norms, gradient-based saliency, and knockout scores) and compare them with standard methods such as LASSO, Random Forest feature importance, Mutual Information, and SVM-RFE on a suite of real and synthetic classification and regression datasets. Using average F1 and $R^2$ scores across three feature-retention levels (20%, 40%, 60%), we find that KAN-based selectors are generally competitive with, and sometimes superior to, classical baselines. In classification, KAN criteria often match or exceed existing methods on multi-class tasks by removing redundant features and capturing nonlinear interactions. In regression, KAN-based scores provide robust performance on noisy and heterogeneous datasets, closely tracking strong ensemble predictors; we also observe characteristic failure modes, such as overly aggressive pruning with an $\ell_1$ criterion. Stability and redundancy analyses further show that KAN-based selectors yield reproducible feature subsets across folds while avoiding unnecessary correlation inflation, ensuring reliable and non-redundant variable selection. Overall, our findings demonstrate that KAN-based feature selection provides a powerful and interpretable alternative to traditional methods, capable of uncovering nonlinear and multivariate feature relevance beyond sparsity or impurity-based measures.

دراسة حديثة تقدم إطارًا لاختيار الميزات في البيانات الجدولية الخاضعة للإشراف باستخدام شبكات كولموغوروف-أرنولد (KAN)، التي تستخدم المنحنيات لاستنتاج درجات أهمية الميزات. تقارن الدراسة معايير الاختيار المستندة إلى KAN بأساليب تقليدية مثل LASSO وRandom Forest، مما يظهر أداءً تنافسيًا في مهام التصنيف والانحدار عبر مجموعات بيانات متنوعة.

Un estudio reciente presenta un marco para la selección de características en datos tabulares supervisados utilizando Redes de Kolmogorov-Arnold (KAN), que emplean splines para derivar puntajes de importancia de características. La investigación compara los criterios de selección basados en KAN con métodos tradicionales como LASSO y Random Forest, demostrando un rendimiento competitivo en tareas de clasificación y regresión en varios conjuntos de datos.

Une étude récente présente un cadre pour la sélection de caractéristiques dans des données tabulaires supervisées utilisant des Réseaux de Kolmogorov-Arnold (KAN), qui utilisent des splines pour dériver des scores d'importance des caractéristiques. La recherche compare les critères de sélection basés sur KAN à des méthodes traditionnelles comme LASSO et Random Forest, montrant une performance compétitive dans des tâches de classification et de régression sur divers ensembles de données.

A recent study introduces a framework for feature selection in supervised tabular data using Kolmogorov-Arnold Networks (KANs), which utilize splines to derive feature importance scores. The research compares KAN-based selection criteria against traditional methods like LASSO and Random Forest, demonstrating competitive performance in classification and regression tasks across various datasets.

Splines-Based Feature Importance in Kolmogorov-Arnold Networks: A Framework for Supervised Tabular Data Dimensionality Reduction

arXiv:2511.13344v3 Announce Type: replace 
Abstract: This paper presents a novel Mixture-of-Experts framework for object detection, incorporating adaptive routing among multiple YOLOv9-T experts to enable dynamic feature specialization and achieve higher mean Average Precision (mAP) and Average Recall (AR) compared to a single YOLOv9-T model.

تم تقديم إطار جديد من Mixture-of-Experts للكشف عن الأجسام، مستفيدًا من التوجيه التكيفي بين عدة خبراء من YOLOv9-T. يهدف هذا النهج إلى تحسين تخصص الميزات، مما يؤدي إلى زيادة دقة المتوسط (mAP) وزيادة الاسترجاع (AR) مقارنة بالطرق التقليدية التي تعتمد على نموذج واحد.

Se ha introducido un nuevo marco de Mixture-of-Experts para la detección de objetos, aprovechando el enrutamiento adaptativo entre múltiples expertos YOLOv9-T. Este enfoque tiene como objetivo mejorar la especialización de características, resultando en una mayor precisión media (mAP) y un mayor recuerdo (AR) en comparación con los métodos tradicionales de modelo único.

Un nouveau cadre Mixture-of-Experts a été introduit pour la détection d'objets, utilisant un routage adaptatif entre plusieurs experts YOLOv9-T. Cette approche vise à améliorer la spécialisation des caractéristiques, entraînant une meilleure précision moyenne (mAP) et un meilleur rappel (AR) par rapport aux méthodes traditionnelles à modèle unique.

A novel Mixture-of-Experts framework has been introduced for object detection, leveraging adaptive routing among multiple YOLOv9-T experts. This approach aims to enhance feature specialization, resulting in improved mean Average Precision (mAP) and Average Recall (AR) compared to traditional single-model methods.

RadioKMoE: Knowledge-Guided Radiomap Estimation with Kolmogorov-Arnold Networks and Mixture-of-Experts

Was this article worth reading? Share it