This paper investigates a novel learning-based sliding mode control (SMC) strategy for a class of nonhomogeneous semi-Markov switching systems with incomplete sojourn-time information. In contrast to existing studies, the sojourn-time probability density functions of semi-Markov kernel are considered to be incompletely known, relaxing the assumption of completely known sojourn-time distributions. The main novelty lies in constructing a novel learning-based SMC scheme to realize the attainment of the quasi-sliding mode, overcoming the difficulty caused by incomplete sojourn-time information. By utilizing the upper bounds of sojourn time for each subsystem mode and constructing the multiple-Lyapunov function, sufficient conditions for mean-square stability are derived for the resulting closed-loop system. Moreover, a recursive sliding mode learning controller is developed to ensure finite-time reachability of the sliding motion and to effectively suppress both parametric uncertainties and switching-induced chattering. The effectiveness and practical applicability of the proposed method are validated through the DC-DC buck converter application.

While existing far memory and disaggregated memory solutions provide a foundation for addressing limitations of single-node memory capacity and inefficient resource allocation in data centers, they predominantly focus on host memory, overlooking the critical demands of GPU-centric workloads. A key bottleneck in scaling GPU memory is the lack of connectivity and interoperability between GPUs, which is exacerbated by their heterogeneity. To bridge this gap, this paper proposes UHM, a unified data transferring and memory pooling scheme for heterogeneous GPU memories. UHM establishes the communication channels between heterogeneous GPU/host memories and leverages double data buffers for pipelined and reliable transfer. Furthermore, UHM unifies both local and remote memories to build a memory pool. The pooling scheme effectively integrates local and remote resources, performs efficient caching management in local memory, and optimizes memory block management for remote memory resources. Evaluation on a heterogeneous GPU cluster demonstrates that UHM significantly reduces the data transfer latency (up to 87.2%), improves the cache hit ratio, reduces runtime memory allocation latency (up to 94.7%), while enhancing the overall memory utilization (24.7%).

Diabetic Retinopathy (DR) represents a significant ophthalmic complication of diabetes, demanding accurate and interpretable diagnostic tools to mitigate vision loss through timely interventions. Existing supervised learning techniques, while successful in general image tasks, face critical limitations in medical contexts due to the high cost and complexity of detailed clinical annotations required for precise DR grading. In response, we propose DR-Reasoning, a novel multimodal vision-language reasoning framework tailored specifically for DR classification tasks. Our method selects the Qwen2.5-VL-7BInstruct model as the base model, and is trained on a real hospital dataset using Low-Rank Adaptation (LoRA) and Group Relative Policy Optimization (GRPO). DR-Reasoning integrates explicit reasoning chains into the diagnostic process, enabling effective capture of subtle pathological features in both fundus images and clinical descriptions. Extensive evaluations demonstrate that DRReasoning achieves state-of-the-art performance, substantially outperforming baseline methods across critical metrics such as AUC (92.79%) and specificity (97.69%).

Superconducting electrodynamic suspension (SC-EDS) train, characterized by its large levitation gap, high lift-to-drag ratio, and self-stabilization features, has significant application potential in high-speed and ultra-high-speed maglev transportation. However, the modeling for the SC-EDS train poses significant challenges in balancing accuracy with computational efficiency. This paper proposes a physics-informed neural network (PINN)-based modeling method for the SC-EDS train. Focusing on the dynamic circuit equations governing the levitation/guidance (L/G) coils and onboard SC coils, the method leverages PINN to rapidly solve for the currents in the L/G coils, thereby enabling real-time computation of electromagnetic forces critical for dynamic modeling of the train. By embedding the dynamic circuit equations into neural networks as physical constraints, the model achieves high-fidelity approximations of the currents while drastically reducing computational costs. This approach not only enhances the feasibility of real-time dynamic modeling but also lays a foundation for stability analysis, system optimization, and advanced control strategies in SC-EDS applications.

As network scale and complexity continue to increase, managing network configurations has become an increasingly challenging task. Existing configuration tools often depend on low-level, abstract intermediate representations, which require users to have substantial technical expertise. This reliance not only increases the learning curve but also heightens the risk of configuration errors. Recent advances in Large Language Models (LLMs) have demonstrated strong potential for automating tasks across various domains. However, their applications to network configuration generation remain limited due to several challenges, including hallucination, restricted context length, and insufficient adaptability to domain-specific requirements. To address these issues, we propose ConfAgent, an advanced network configuration generation system powered by a multi-model intelligent agent. ConfAgent comprises four key components: a conflict detector, an information extractor, a routing algorithm coder, and a formal synthesizer. These components collaborate to accurately interpret complex configuration intents, detect potential conflicts, and generate robust code and network configurations through intuitive natural language interactions. Extensive experiments conducted on the NetConfEval benchmark demonstrate that ConfAgent consistently outperforms existing state-of-the-art methods by margins ranging from 36 % to 100 %, particularly excelling in configuration tasks for large-scale network topologies.

Network traffic classification plays a crucial role in a wide range of applications, e.g., Quality of Service (QoS) enhancement, resource management, and network security. However, the widespread adoption of encryption protocols (e.g., SSL/TLS) and the emergence of anonymous communication systems (e.g., Tor) have introduced significant challenges due to the presence of complex and varied network noise. Although considerable effort has been made to improve the robustness of the traffic classification, the performances of existing state-of-theart methods cannot be guaranteed in the presence of a mixture of noises, and are not stable in different application scenarios. In this paper, we innovate in proposing a network traffic classification method based on MoA (Mixture of Agents), namely MOTA. By leveraging a light-weight MoA architecture, MOTA efficiently fine-tunes mainstream Large Language Models (LLMs) to adapt to different application scenarios of traffic classification, and fully exploits the collaboration of the LLMs to ensure the robustness against mixed noises. Our extensive experiments show that, the classification accuracy is$\geq 99 {\%}$across multiple public datasets injected with mixed noises, significantly outperforming existing SOTA methods. Moreover, despite incorporating multiple LLMs, MOTA maintains millisecond-level inference latency on a server equipped with four NVIDIA GeForce RTX 4090 GPUs, owing to its lightweight design.

The rapid development of railways worldwide has generated significant attention toward high-speed train (HST) communications. Disruptive technologies for the sixth-generation network, such as cell-free massive multiple-input–multiple-output (mMIMO) and reconfigurable intelligent surface (RIS), are set to substantially enhance the data transmission rate in wireless communication systems. In this article, we leverage the potential benefits of cell-free mMIMO and irregular RIS to achieve the ultimate performance benchmark in HST networks. Specifically, we investigate a cell-free mMIMO system supported by irregular RIS with many low-cost reflecting elements, aiming to improve the weighted sum-rate (WSR) in HST communications. The deployment of irregular RIS allows signals to be transmitted through reflected links with additional degrees of freedom, and thus, a more diverse transmission is facilitated to enhance the received signal power and the system capacity. A joint optimization problem is then formulated to maximize the WSR, while satisfying the constraints on access point power. To address the problem characterized by nonconvexity and high complexity, an alternating optimization algorithm is proposed to iteratively optimize the RIS topology, the transmit beamforming, and the phase shifts. The corresponding subproblems are tackled by exploiting improved tabu search, projected subgradient, and nonmonotone accelerated proximal gradient methods, respectively. Extensive simulations confirm the significant advantages of irregular RIS configurations on the WSR.