arXiv:2510.24546v1 Announce Type: cross 
Abstract: Despite the popularity of reinforcement learning (RL) in wireless networks, existing approaches that rely on model-free RL (MFRL) and model-based RL (MBRL) are data inefficient and short-sighted. Such RL-based solutions cannot generalize to novel network states since they capture only statistical patterns rather than the underlying physics and logic from wireless data. These limitations become particularly challenging in complex wireless networks with high dynamics and long-term planning requirements. To address these limitations, in this paper, a novel dual-mind world model-based learning framework is proposed with the goal of optimizing completeness-weighted age of information (CAoI) in a challenging mmWave V2X scenario. Inspired by cognitive psychology, the proposed dual-mind world model encompasses a pattern-driven System 1 component and a logic-driven System 2 component to learn dynamics and logic of the wireless network, and to provide long-term link scheduling over reliable imagined trajectories. Link scheduling is learned through end-to-end differentiable imagined trajectories with logical consistency over an extended horizon rather than relying on wireless data obtained from environment interactions. Moreover, through imagination rollouts, the proposed world model can jointly reason network states and plan link scheduling. During intervals without observations, the proposed method remains capable of making efficient decisions. Extensive experiments are conducted on a realistic simulator based on Sionna with real-world physical channel, ray-tracing, and scene objects with material properties. Simulation results show that the proposed world model achieves a significant improvement in data efficiency and achieves strong generalization and adaptation to unseen environments, compared to the state-of-the-art RL baselines, and the world model approach with only System 1.

إطار عمل جديد يسمى نماذج العالم ذات العقلين يهدف إلى تحسين التعلم في الشبكات اللاسلكية الديناميكية من خلال معالجة قيود أساليب التعلم المعزز الحالية. غالبًا ما تكافح الأساليب التقليدية، سواء كانت بدون نموذج أو قائمة على نموذج، مع كفاءة البيانات وتفشل في التكيف مع حالات الشبكة الجديدة، حيث تعتمد بشكل مفرط على الأنماط الإحصائية. يسعى هذا الإطار إلى دمج فهم أعمق للفيزياء والمنطق الكامن وراء البيانات اللاسلكية، مما قد يؤدي إلى حلول أكثر فعالية في بيئات الشبكة المعقدة.

Un nuevo marco llamado Dual-Mind World Models busca mejorar el aprendizaje en redes inalámbricas dinámicas al abordar las limitaciones de los métodos actuales de aprendizaje por refuerzo. Los enfoques tradicionales, tanto sin modelo como basados en modelo, a menudo luchan con la eficiencia de los datos y no logran adaptarse a nuevos estados de red, confiando demasiado en patrones estadísticos. Este marco busca incorporar una comprensión más profunda de la física y la lógica subyacentes de los datos inalámbricos, lo que podría llevar a soluciones más efectivas en entornos de red complejos.

Un nouveau cadre appelé Dual-Mind World Models vise à améliorer l'apprentissage dans les réseaux sans fil dynamiques en abordant les limites des méthodes d'apprentissage par renforcement actuelles. Les approches traditionnelles sans modèle et basées sur un modèle ont souvent du mal avec l'efficacité des données et échouent à s'adapter à de nouveaux états de réseau, s'appuyant trop sur des modèles statistiques. Ce cadre cherche à intégrer une compréhension plus profonde de la physique et de la logique sous-jacentes des données sans fil, ce qui pourrait conduire à des solutions plus efficaces dans des environnements de réseau complexes.

A new framework called Dual-Mind World Models aims to improve learning in dynamic wireless networks by addressing the limitations of current reinforcement learning methods. Traditional model-free and model-based approaches often struggle with data efficiency and fail to adapt to new network states, relying too heavily on statistical patterns. This framework seeks to incorporate a deeper understanding of the underlying physics and logic of wireless data, which could lead to more effective solutions in complex network environments.

Dual-Mind World Models: A General Framework for Learning in Dynamic Wireless Networks

arXiv:2511.17228v1 Announce Type: cross 
Abstract: Artificial intelligence in dynamic, real-world environments requires the capacity for continual learning. However, standard deep learning suffers from a fundamental issue: loss of plasticity, in which networks gradually lose their ability to learn from new data. Here we show that quantum learning models naturally overcome this limitation, preserving plasticity over long timescales. We demonstrate this advantage systematically across a broad spectrum of tasks from multiple learning paradigms, including supervised learning and reinforcement learning, and diverse data modalities, from classical high-dimensional images to quantum-native datasets. Although classical models exhibit performance degradation correlated with unbounded weight and gradient growth, quantum neural networks maintain consistent learning capabilities regardless of the data or task. We identify the origin of the advantage as the intrinsic physical constraints of quantum models. Unlike classical networks where unbounded weight growth leads to landscape ruggedness or saturation, the unitary constraints confine the optimization to a compact manifold. Our results suggest that the utility of quantum computing in machine learning extends beyond potential speedups, offering a robust pathway for building adaptive artificial intelligence and lifelong learners.

أظهرت التطورات الأخيرة في نماذج التعلم الكمومي قدرتها على الحفاظ على اللدونة في بيئات التعلم المستمر، مما يعالج قيدًا كبيرًا موجودًا في أنظمة التعلم العميق التقليدية. تظهر هذه النماذج قدرات تعلم متسقة عبر مجموعة متنوعة من المهام وأنواع البيانات، بما في ذلك التعلم المراقب والتعلم المعزز.

Los avances recientes en modelos de aprendizaje cuántico han demostrado su capacidad para mantener la plasticidad en entornos de aprendizaje continuo, abordando una limitación significativa que se encuentra en los sistemas de aprendizaje profundo tradicionales. Estos modelos muestran capacidades de aprendizaje consistentes en diversas tareas y tipos de datos, incluidos el aprendizaje supervisado y el aprendizaje por refuerzo.

Des avancées récentes dans les modèles d'apprentissage quantique ont démontré leur capacité à maintenir la plasticité dans des environnements d'apprentissage continu, répondant à une limitation significative des systèmes d'apprentissage profond traditionnels. Ces modèles montrent des capacités d'apprentissage cohérentes à travers diverses tâches et types de données, y compris l'apprentissage supervisé et l'apprentissage par renforcement.

Recent advancements in quantum learning models have demonstrated their ability to maintain plasticity in continual learning environments, addressing a significant limitation found in traditional deep learning systems. These models show consistent learning capabilities across various tasks and data types, including supervised and reinforcement learning.

Dual-Mind World Models: A General Framework for Learning in Dynamic Wireless Networks

Was this article worth reading? Share it

Chattermate

Metaflow AI

Invent