Bin Zhang

Mobile Edge Computing (MEC) enables low-latency, energy-efficient offloading in next-generation wireless networks, but its performance degrades when direct links are blocked in dense urban areas. To address this, recent studies integrate Unmanned Aerial Vehicles (UAVs) and Reconfigurable Intelligent Surfaces (RIS) to enhance coverage and efficiency through UAV mobility and RIS-assisted virtual line-of-sight paths. However, existing centralized optimization approaches suffer from high complexity and communication overhead, making real-time coordination difficult in dynamic multi-UAV settings. To overcome these challenges, we propose a decentralized, model-based reinforcement learning (RL) framework for UAV-RIS-MEC systems, where each UAV learns trajectory, RIS configuration, task offloading, and resource allocation policies using local observations and limited $\kappa$-hop neighbor communication. By employing short-horizon branched rollouts with Proximal Policy Optimization (PPO), the proposed method achieves stable and sample-efficient learning with theoretical convergence guarantees. Simulation results show notable gains in throughput and energy efficiency over centralized and conventional decentralized baselines, demonstrating the effectiveness and scalability of the proposed approach for future MEC deployments.

Aiming at the core challenges of data shortage and high labeling cost in the infrared small target detection task, this paper puts forward an innovative method of integrating generative adversarial network and semi-supervised learning. By designing the small target generative adversarial network (STGAN) based on StyleGAN2, combining the self-attention mechanism and the comparative learning loss, the quality and diversity of generated data are effectively improved. At the same time, the knowledge distillation framework is used to optimize the student model by using the pseudo-tags generated by the teacher model, which significantly improves the detection performance in small sample scenes. The experimental results show that the nIoU (1% labeled data) of 0.709 is achieved by the proposed method on SIRST data set, and the FID index is reduced to 0.45, both reaching the current optimal level. The ablation experiment further verified the key roles of detection loss, self-attention module and REP-PAN feature fusion strategy, and provided a new idea for data expansion and model optimization in the field of small target detection.

In the domain of automatic parking systems, parking space detection and localization represent fundamental challenges that must be addressed. As a core research focus within the field of intelligent automatic parking, they constitute the essential prerequisite for the realization of fully autonomous parking. Accurate and effective detection of parking spaces is still the core problem that needs to be solved in automatic parking systems. In this study, building upon existing public parking space datasets, a comprehensive panoramic parking space dataset named PSEX (Parking Slot Extended) with complex environmental diversity was constructed by integrating the concept of GAN (Generative Adversarial Network)-based image style transfer. Meanwhile, an improved algorithm based on PP-Yoloe (Paddle-Paddle Yoloe) is used to detect the state (free or occupied) and angle (T-shaped or L-shaped) of the parking space in real-time. For the many and small labels of the parking space, the ResSpp in it is replaced by the ResSimSppf module, the SimSppf structure is introduced at the neck end, and Silu is replaced by Relu in the basic structure of the CBS (Conv-BN-SiLU), and finally an auxiliary detector head is added at the prediction head. Experimental results show that the proposed SimSppf_mepre-Yoloe model achieves an average improvement of 4.5% in mAP50 and 2.95% in mAP50:95 over the baseline PP-Yoloe across various parking space detection tasks. In terms of efficiency, the model maintains comparable inference latency with the baseline, reaching up to 33.7 FPS on the Jetson AGX Xavier platform under TensorRT optimization. And the improved enhancement algorithm can greatly enrich the diversity of parking space data. These results demonstrate that the proposed model achieves a better balance between detection accuracy and real-time performance, making it suitable for deployment in intelligent vehicle and robotic perception systems.

This paper presents the development of an advanced servo system for gluing operations, which integrates a dosing cylinder enhanced by reinforcement learning algorithms to optimize efficiency and accuracy. The system consists of an upper computer, a programmable logic controller (PLC), a main controller (featuring a core board and an IO board), and a dosing cylinder equipped with a servo motor and a glue gun. A comprehensive system model was developed by 1 analyzing the interactions among mechanical, hydraulic, and servo motor subsystems. The implementation of an enhanced Deep Deterministic Policy Gradient (DDPG) reinforcement learning algorithm, augmented with Model Predictive Control (MPC), enables precise control over voltage inputs and glue gun temperature adjustments. The MPC component predicts future states of the system based on the current state and actions, refining the action selection process by considering the long-term effects of actions. Rigorous simulation testing of the enhanced DDPG algorithm, alongside other algorithms, in configurations involving single and dual cylinder setups, validates its effectiveness, demonstrating significant enhancements in the automated control of the gluing process. The real-world experimental results indicate high positioning accuracy, consistent glue dispensing , and overall stability, making this system suitable for industrial applications requiring precise and reliable gluing.

The exponentially growing content on the Internet includes online publications, scientific news, and other expert websites. It brings a formidable challenge to extracting pertinent answers from such vast information. We proposed a deep neural network model based on BiLSTM(Bi-directional Long Short-Term Memory) and ALBERT(A Lite BERT for Self-Supervised Learning of Language Representations) for Chinese question-answering in the scientific context. Our model significantly enhances the generalization capabilities of the transformer-based model for question answering. The character extraction and word embedding modules are designed for tackling intricate science-related queries, swiftly assimilating knowledge from cutting-edge scientific literature, and contributing to constructing a comprehensive scientific knowledge graph. Our model is characterized by compactness, swift execution, and satisfactory accuracy. It has been meticulously fine-tuned, emphasizing multi-sentence coherence augmented by attention mechanisms, thereby ensuring robust scalability. Empirical evaluations on the LCQMC and XNLI datasets demonstrate that our approach surpasses the performance results of BERT and ALBERT, showing the potential for large-scale Chinese question-matching problems.

Traffic flow counting is an object detection problem. YOLO (" You Only Look Once ") is a popular object detection network. Spiking-YOLO converts the YOLO network from ANN(artificial neural network) to SNN (spiking neural network), which consumes less power than YOLO. However, Spiking-YOLO is difficult to train and has poor accuracy because of overfitting. Dropout and DropConnect are two methods to regularize neural network models and have produced cutting-edge outcomes in many benchmarks. The two drop approaches are first applied to Spiking-YOLO. Considering that a higher membrane potential of a biological neuron implies a higher probability of neural activation, we propose three adaptive drop algorithms—Spiking-YOLO with adaptive Dropout, Spiking-YOLO with adaptive DropConnect, and Spiking-YOLO with group adaptive Drop, which adaptively adjusts the keep probability for training SNNs. The experimental results show that the adaptive algorithms have fast convergence speed with high detection accuracy and mAP value, and work well in the case of high-speed streaming video, making them suitable for online learning scenarios.

Object detection techniques are used to determine object categories and localization in images. Different tasks have different requirements for detection efficiency and accuracy, which leads to high efficiency one-stage object detection methods and high-precision two-stage object detection methods. As the main category of difficult detection objects, small objects have the characteristics of few pixels, wide distribution, and complex background. Many serious tasks in special environments, such as industrial defect detection, large-scale pedestrian statistics, remote sensing image recognition, etc. require the model to have high accuracy for the detection of difficult objects. Aiming at the problem of information loss of small objects in deep network features, this paper proposes a feature levels selection method for backbone network based on effective receptive fields. By moving down the feature level of the backbone network, the size of the effective receptive field corresponding to the selected hierarchical features is matched with the scale distribution of the small objects to be detected to obtain the best small object detection performance. According to the effective receptive field size (ERF), the appropriate Shallow Feature Selection (SFS) is selected to participate in the subsequent detection process to ensure that accurate and sufficient small target information is extracted from the backbone network.Based on one-stage and two-stage object detection methods, this paper proposes improvement measures for small object detection from the data level, network level and training level, and designs experiments and validates them.