检索结果-内蒙古大学图书馆

MIS-YOLOv8: An Improved Algorithm for Detecting Small Objects in UAV Aerial Photography Based on YOLOv8

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Tao, Sun Shengqi, Yang Haiying, Liu Jason, Gu Lixia, Deng Lida, Liu Qilu Univ Technol Shandong Acad Sci Sch Informat & Automat Engn Jinan 250353 Peoples R China Dalhousie Univ Sch Elect & Comp Engn Halifax NS B3J 1Z1 Canada Shandong Runyi Intelligent Technol Co Ltd Jinan 250002 Peoples R China

Small objects' detection from a drone's perspective has always been a challenging issue in the field of object detection. To address the problems of low recognition accuracy and information loss in small object detection, this article proposed MIS-YOLOv8 algorithm, primarily aimed at resolving the issue of small object loss during the detection process of the classic YOLOv8s algorithm. First, a multilevel feature extraction (MFE) module was designed for enriching the feature representation capabilities capable of extracting objects from different scales. Second, a small object detection mechanism was incorporated for improving the detection ability. Finally, the integration of depthwise atrous flexible convolutions is introduced, enabling a rich capture of information from spatial to depth dimensions, thereby reducing the loss of small objects. The improved MIS-YOLOv8 algorithm validation was conducted on the VisDrone2019 dataset, where MIS-YOLOv8 demonstrated a 9% and 6.2% increase in mAP@0.5 and mAP@0.5:0.95, respectively, compared with YOLOv8s. The experimental results indicated that the improved model exhibits superior performance in small object detection for drones.

关键词： Feature extraction Computational modeling YOLO Training Accuracy Computational efficiency Drones inference algorithms Deep learning Autonomous aerial vehicles small object detection unmanned aerial vehicle (UAV) YOLOv8s

来源：评论

学校读者我要写书评

暂无评论

Counterfactual Data Generation Method for Fault Diagnosis of Complex Electromechanical Systems

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Wang, Chong Liu, Jie Cao, Junwei Chen, Xi Chen, Lihua Ji, Yindong Tsinghua Univ Dept Automat Beijing 100190 Peoples R China Beihang Univ Sch Reliabil & Syst Engn Beijing 100191 Peoples R China Tsinghua Univ Beijing Natl Res Ctr Informat Sci & Technol Beijing 100190 Peoples R China Southwest Jiaotong Univ Sch Informat Sci & Technol Chengdu 610032 Sichuan Peoples R China

The high reliability and maintainability of complex electromechanical systems make the collection of representative fault data difficult in routine operations. This issue poses significant challenges to data-driven models, hindering their ability to accurately capture system fault mechanisms. Generating simulated fault samples is a popular and effective approach to augment fault data. However, state-of-the-art data generation methods can only produce synthetic data with respect to the distribution of the collected data and may violate the actual fault mechanisms. To address this, this article proposes a counterfactual data generation method grounded in causality, aiming to simulate the intrinsic generation process of monitoring data and thereby obtain counterfactual data that conforms to the system's fault mechanisms. In the proposed method, a priori-constrained causal discovery method is designed to uncover the causalities among the monitoring variables in complex electromechanical systems. A graph decoupling network is designed to disentangle causal mechanisms and extract decoupled features representing different uncertainty sources of the monitoring data. Finally, a causal-based generative adversarial network (CGAN) is proposed to generate counterfactual monitoring data that satisfies the mined causalities. The experimental results show that the generalizability and stability of fault diagnosis models can be enhanced with the counterfactual data.

关键词： Monitoring Electromechanical systems Data collection Cause effect analysis Fault diagnosis inference algorithms Training Roads Generative adversarial networks Data models Causality counterfactual data generation decoupling causal representation electromechanical systems fault diagnosis

来源：评论

学校读者我要写书评

暂无评论

Hybrid Attention for Robust RGB-T Pedestrian Detection in Real-World Conditions

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2025年第1期10卷 319-326页

作者： Rathinam, Arunkumar Pauly, Leo Shabayek, Abd El Rahman Rharbaoui, Wassim Kacem, Anis Gaudilliere, Vincent Aouada, Djamila Univ Luxembourg Interdisciplinary Ctr Secur Reliabil & Trust SnT L-1855 Kirchberg Luxembourg Univ Luxembourg Interdisciplinary Ctr Secur Reliabil & Trust SnT L-4365 Esch Sur Alzette Luxembourg Univ Poitiers XLIM Inst F-87060 Limoges France Univ Lorraine CNRS Inria Loria F-54000 Nancy France

Multispectral pedestrian detection has gained significant attention in recent years, particularly in autonomous driving applications. To address the challenges posed by adversarial illumination conditions, the combination of thermal and visible images has demonstrated its advantages. However, existing fusion methods rely on the critical assumption that the RGB-Thermal (RGB-T) image pairs are fully overlapping. These assumptions often do not hold in real-world applications, where only partial overlap between images can occur due to sensors configuration. Moreover, sensor failure can cause loss of information in one modality. In this letter, we propose a novel module called the Hybrid Attention (HA) mechanism as our main contribution to mitigate performance degradation caused by partial overlap and sensor failure, i.e. when at least part of the scene is acquired by only one sensor. We propose an improved RGB-T fusion algorithm, robust against partial overlap and sensor failure encountered during inference in real-world applications. We also leverage a mobile-friendly backbone to cope with resource constraints in embedded systems. We conducted experiments by simulating various partial overlap and sensor failure scenarios to evaluate the performance of our proposed method. The results demonstrate that our approach outperforms state-of-the-art methods, showcasing its superiority in handling real-world challenges.

关键词： Sensors Pedestrians Feature extraction Thermal sensors Cameras Robustness inference algorithms Sensor fusion Lighting Image sensors Deep learning for visual perception multi-modal perception for HRI sensor fusion human detection and tracking computer vision for transportation

来源：评论

学校读者我要写书评

暂无评论

Lightweight Strategies for Decision-Making of Autonomous Vehicles in Lane Change Scenarios Based on Deep Reinforcement Learning

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2025年第5期26卷 7245-7261页

作者： Li, Guofa Yan, Jun Qiu, Yifan Li, Qingkun Li, Jie Li, Shengbo Eben Green, Paul Chongqing Univ Coll Mech & Vehicle Engn Chongqing 400044 Peoples R China Shenzhen Univ Inst Human Factors & Ergon Coll Mechatron & Control Engn Shenzhen 518060 Peoples R China Chinese Acad Sci Inst Software Beijing Key Lab Human Comp Interact Beijing 100190 Peoples R China Tsinghua Univ Sch Vehicle & Mobil Beijing 100084 Peoples R China Univ Michigan Transportat Res Inst UMTRI Ann Arbor MI 48109 USA

High-performance vision-based decision-making networks are often limited by hardware capabilities in practical applications. To address this challenge, this study proposes lightweight optimization strategies for decision-making models from the aspects of parameter size, training memory usage, and inference speed. Specifically, an innovative solution is proposed to achieve lightweight parameters. The Video Swin Transformer is employed to simultaneously extract temporal and spatial features, with the network trained using a Prioritized Replay Deep Q-Network (PRDQN) that incorporates risk assessment. To further reduce training memory usage, the Q-target network in PRDQN is removed, and the mellowmax operator is integrated to enhance the training process, resulting in the PRDeepMellow Swin Transformer. After analyzing the inference speed problems encountered by the algorithm in practical applications, the vanilla self-attention is replaced by a linear self-attention based on double softmax, namely Double Softmax Linear Video Swin Transformer (DSLVS Transformer) which improves the inference speed for long sequences. The proposed methods were evaluated across three high-speed lane change scenarios (a static scenario, a dynamic scenario, and a randomly changing scenario). Experimental results demonstrate that the proposed methods can still maintain excellent decision performance after the corresponding lightweight optimizations.

关键词： Transformers Training Decision making Autonomous vehicles Feature extraction Computational modeling Q-learning Electronic mail inference algorithms Hardware Driving safety autonomous vehicles lane change decision making reinforcement learning light-weight

来源：评论

学校读者我要写书评

暂无评论

Quantifying Privacy Risks of Behavioral Semantics in Mobile Communication Services

引用

IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY 2025年 20卷 1908-1923页

作者： Qiu, Guoying Bai, Tiecheng Tang, Guoming Guo, Deke Li, Chuandong Gan, Yan Zhou, Baoping Shen, Yulong Tarim Univ Coll Cyber Secur Xinjiang 843300 Peoples R China Hong Kong Univ Sci & Technol Guangzhou Data Sci & Analyt DSA Thrust Informat Hub Guangzhou 510220 Guangdong Peoples R China Sun Yat Sen Univ Sch Engn Guangzhou 510220 Guangdong Peoples R China Southwest Univ Coll Elect & Informat Engn Chongqing 400715 Peoples R China Chongqing Univ Coll Comp Sci Chongqing 400044 Peoples R China Xidian Univ Coll Comp Sci & Technol Xian 710071 Peoples R China Xidian Univ Shaanxi Key Lab Network & Syst Secur Xian 710071 Peoples R China

Location-based mobile services, while improving user daily life, also raise significant privacy concerns in the sharing of location data. These trajectories indicate users' traveling behavioural traces with rich semantics derived from open-source information. Behavioral-semantic analysis reveals users' travelling motivations and underlying behavioral patterns. It contributes to attackers launching inferential attacks for behavior prediction, identity identification, or other privacy invasions, even when the location data is protected. It remains open to the issues of behavioral-semantic privacy-risk quantification and privacy-protection evaluation. This paper aims to reveal such semantic privacy risks of user behaviors arising from the publication of location trajectories in mobile scenarios. We formalize user semantic-mobility process to analyze his underlying behavior patterns. Then, we design semantic inference algorithms conditional on the released trajectory to reason about the observation-based likelihood of the user's actual staying and transfer behaviours and behavioural-trace tracking. Extensive experiments with real-world data demonstrate their performance on inference accuracy and semantic similarity, offering a quantification criterion for deploying mobile privacy protection.

关键词： Semantics Privacy Trajectory Protection Motion pictures Data privacy inference algorithms Hidden Markov models Stochastic processes Accuracy Mobile service location privacy protection behavior semantics privacy-risk evaluation & quantification

来源：评论

学校读者我要写书评

暂无评论

AB-VIP: Attack Method Using Brightness Information for Visibility Improvement of Adversarial Perturbations on the NMS Algorithm

引用

IEEE ACCESS 2025年 13卷 66065-66077页

作者： Kim, Gwang-Nam Lee, Han-Ju Lee, Han-Jin Jung, Hyeon-Jin Choi, Seok-Hwan Yonsei Univ Dept Comp Sci Wonju 26493 South Korea Yonsei Univ Div Software Wonju 26493 South Korea

As with classification models, object detection models are vulnerable to adversarial attacks. In particular, adversarial attacks on key components of object detection models such as Region Proposal Network (RPN) and Non-Maximum Suppression (NMS) algorithms have recently been highlighted. As a representative adversarial attack method targeting the NMS algorithm, PhantomSponges disrupts the normal operation of the NMS algorithm by using universal perturbations. However, since PhantomSponges focuses solely on the attack success rate, the adversarial examples it generates are visually noticeable. In this paper, we propose a simple yet effective attack method called Attack method using Brightness Information for Visibility Improvement(AB-VIP), to address the limitation of adversarial attack methods targeting the NMS algorithm. To improve the invisibility of adversarial examples, the proposed method adjusts universal perturbation using the brightness information of each target image. From the experimental results under various conditions, we demonstrate that the proposed method improves the invisibility of the existing NMS-targeted attack method while maintaining a high attack success rate. When applying the proposed method, the SSIM metric improved by 13.23%, while the L2 metric decreased by 66.23%.

关键词： Perturbation methods Object detection Brightness Proposals Classification algorithms Artificial neural networks Noise Measurement inference algorithms Image classification Adversarial attack cyber security NMS-algorithm object detection universal perturbation

来源：评论

学校读者我要写书评

暂无评论

Meta-Learning Based Indoor Localization: A Hierarchical Framework

引用

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY 2025年第4期74卷 6697-6702页

作者： Yu, Haiyao She, Changyang Hu, Yunkai Geng, Zhaoquan Vucetic, Branka Li, Yonghui Univ Sydney Sch Elect & Comp Engn Sydney NSW 2006 Australia Univ Sydney Sch Elect & Informat Engn Sydney NSW 2006 Australia Harbin Inst Technol Shenzhen Sch Elect & Informat Engn Shenzhen 518055 Peoples R China

Machine learning (ML) has been considered a promising approach for indoor localization. Nevertheless, existing ML approaches require a large amount of computational resources and data samples in the new scenarios or for different types of mobile devices (MD). To address the issues, we first propose a hierarchical framework that divides the deep neural network (DNN) for localization into two parts and deploys them on the server and MD, respectively. The on-device part can estimate a coarse location of the MD with local information, and the other part can improve the accuracy with global information on the server. Then, we develop a meta-learning-based training algorithm that trains the DNN in different scenarios or with different types of MDs. Then, it is implemented in unseen scenarios or on new types of MDs, where a small number of data samples are available for fine-tuning the DNN. By implementing a learning-based localization algorithm in the hierarchical framework, the root mean square error (RMSE) achieved by the on-device DNN is 1.18 m and the inference time is less than 20 ms. With the help of global information at the server, the RMSE is reduced to 0.6 m.

关键词： Location awareness Servers Accuracy Training Antenna measurements Metalearning Artificial neural networks Semiconductor device measurement Wireless fidelity inference algorithms Deep vision transformer graph neural networks indoor localization meta-learning

来源：评论

学校读者我要写书评

暂无评论

DeepTrace: Learning to Optimize Contact Tracing in Epidemic Networks With Graph Neural Networks

引用

IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS 2025年 11卷 97-113页

作者： Tan, Chee Wei Yu, Pei-Duo Chen, Siya Poor, H. Vincent Nanyang Technol Univ Singapore 639798 Singapore Chung Yuan Christian Univ Dept Appl Math Taoyuan City 32023 Taiwan City Univ Hong Kong Dept Comp Sci Hong Kong Peoples R China Princeton Univ Dept Elect & Comp Engn Princeton NJ 08544 USA

Digital contact tracing aims to curb epidemics by identifying and mitigating public health emergencies through technology. Backward contact tracing, which tracks the sources of infection, proved crucial in places like Japan for identifying COVID-19 infections from superspreading events. This paper presents a novel perspective on digital contact tracing by modeling it as an online graph exploration problem, framing forward and backward tracing strategies as maximum-likelihood estimation tasks that leverage iterative sampling of epidemic network data. The challenge lies in the combinatorial complexity and rapid spread of infections. We introduce DeepTrace, an algorithm based on a Graph Neural Network that iteratively updates its estimations as new contact tracing data is collected, learning to optimize the maximum likelihood estimation by utilizing topological features to accelerate learning and improve convergence. The contact tracing process combines either BFS or DFS to expand the network and trace the infection source, ensuring efficient real-time exploration. Additionally, the GNN model is fine-tuned through a two-phase approach: pre-training with synthetic networks to approximate likelihood probabilities and fine-tuning with high-quality data to refine the model. Using COVID-19 variant data, we illustrate that DeepTrace surpasses current methods in identifying superspreaders, providing a robust basis for a scalable digital contact tracing strategy.

关键词： COVID-19 Resistance Maximum likelihood estimation Pandemics Contact tracing Graph neural networks Data models Real-time systems inference algorithms Topology Digital contact tracing contagion source detection maximum likelihood estimation graph neural network online graph exploration

来源：评论

学校读者我要写书评

暂无评论

Communication-Efficient Distributed Bayesian Federated Learning Over Arbitrary Graphs

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2025年 73卷 1351-1366页

作者： Wang, Sihua Guo, Huayan Zhu, Xu Yin, Changchuan Lau, Vincent K. N. Beijing Univ Posts & Telecommun Beijing Lab Adv Informat Network Beijing 100876 Peoples R China Beijing Univ Posts & Telecommun Beijing Key Lab Network Syst Architecture & Conver Beijing 100876 Peoples R China Hong Kong Univ Sci & Technol Dept Elect & Commun Engn Hong Kong 999077 Peoples R China Harbin Inst Technol Sch Informat Sci & Technol Shenzhen 518055 Peoples R China Guangdong Prov Key Lab Aerosp Commun & Networking Shenzhen 518055 Peoples R China Shenzhen Municipal Key Lab AIoT Commun Shenzhen 518055 Peoples R China

This paper investigates a fully distributed federated learning (FL) problem, in which each device is restricted to only utilize its local dataset and the information received from its adjacent devices that are defined in a communication graph to update the local model weights for minimizing the global loss function. To incorporate the communication graph constraint into the joint posterior distribution, we exploit the fact that the model weights on each device is a function of its local likelihood and local prior and then, the connectivity between adjacent devices is modeled by a Dirac Delta distribution. In this way, the joint distribution can be factorized naturally by a factor graph. Based on the Dirac Delta-based factor graph, we propose a novel distributed approximate Bayesian inference algorithm that combines loopy belief propagation (LBP) and variational Bayesian inference (VBI) for distributed FL. Specifically, VBI is used to approximate the non-Gaussian marginal posterior as a Gaussian distribution in local training process and then, the global training process resembles Gaussian LBP where only the mean and variance are passed among adjacent devices. Furthermore, we propose a new damping factor design according to the communication graph topology to mitigate the potential divergence and achieve consensus convergence. Simulation results verify that the proposed solution achieves faster convergence speed with better performance than baselines.

关键词： Bayes methods Training Convergence inference algorithms Damping Brain modeling Approximation algorithms Optimization Signal processing algorithms Federated learning Bayesian federated learning decentralized consensus optimization message passing

来源：评论

学校读者我要写书评

暂无评论

Interacting Large Language Model Agents Bayesian Social Learning Based Interpretable Models

引用

IEEE ACCESS 2025年 13卷 25465-25504页

作者： Jain, Adit Krishnamurthy, Vikram Cornell Univ Sch Elect & Comp Engn Ithaca NY 14853 USA

This paper discusses the theory and algorithms for interacting large language model agents (LLMAs) using methods from statistical signal processing and microeconomics. While both fields are mature, their application to decision-making involving interacting LLMAs remains unexplored. Motivated by Bayesian sentiment analysis on online platforms, we construct interpretable models and stochastic control algorithms that enable LLMAs to interact and perform Bayesian inference. Because interacting LLMAs learn from both prior decisions and external inputs, they can exhibit bias and herding behavior. Thus, developing interpretable models and stochastic control algorithms is essential to understand and mitigate these behaviors. This paper has three main results. First, we show using Bayesian revealed preferences from microeconomics that an individual LLMA satisfies the necessary and sufficient conditions for rationally inattentive (bounded rationality) Bayesian utility maximization and, given an observation, the LLMA chooses an action that maximizes a regularized utility. Second, we utilize Bayesian social learning to construct interpretable models for LLMAs that interact sequentially with each other and the environment while performing Bayesian inference. Our proposed models capture the herding behavior exhibited by interacting LLMAs. Third, we propose a stochastic control framework to delay herding and improve state estimation accuracy under two settings: 1) centrally controlled LLMAs and 2) autonomous LLMAs with incentives. Throughout the paper, we numerically demonstrate the effectiveness of our methods on real datasets for hate speech classification and product quality assessment, using open-source models like LLaMA and Mistral and closed-source models like ChatGPT. The main takeaway of this paper, based on substantial empirical analysis and mathematical formalism, is that LLMAs act as rationally bounded Bayesian agents that exhibit social learning when interacting. Tradi

关键词： Bayes methods Stochastic processes Signal processing algorithms Signal processing Microeconomics inference algorithms Engines Data models Numerical models Bayesian social learning large language models Bayesian revealed preferences structural results optimal stopping POMDPs self-attention rational inattention model collapse model collapse

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：