检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Zhuo, Wei Tang, Zhiyue Xue, Wufeng Ding, Hao Shen, Linlin School of Artificial Intelligence Shenzhen University Shenzhen518060 China National Engineering Laboratory of Big Data System Computing Technology Shenzhen University China School of Biomedical Engineering Shenzhen University Medical School Shenzhen University China School of Computer Science and Software Engineering Shenzhen University China

Few-shot semantic segmentation has gained increasing interest due to its generalization capability, i.e., segmenting pixels of novel classes requiring only a few annotated images. Prior work has focused on meta-learning for support-query matching, with extensive development in both prototype-based and aggregation-based methods. To address data scarcity, recent approaches have turned to foundation models to enhance representation transferability for novel class segmentation. Among them, a hybrid dual-modal framework including both DINOv2 and SAM has garnered attention due to their complementary capabilities. We wonder "can we build a unified model with knowledge from both foundation models?" To this end, we propose FS-DINO, with only DINOv2’s encoder and a lightweight segmenter. The segmenter features a bottleneck adapter, a meta-visual prompt generator based on dense similarities and semantic embeddings, and a decoder. Through coarse-to-fine cross-model distillation, we effectively integrate SAM’s knowledge into our lightweight segmenter, which can be further enhanced by 4D correlation mining on support-query pairs. Extensive experiments on COCO-20i, PASCAL-5i, and FSS-1000 demonstrate the effectiveness and superiority of our method. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Pedestrian Crossing Intention Prediction via Progressive Multimodal Token Fusion for Autonomous Driving

引用

IEEE Transactions on Intelligent Transportation Systems 2025年

作者： Chen, Xiaobo Xu, Wei Zhang, Shilin Cai, Yingfeng Shandong Technology and Business University School of Computer Science and Technology Yantai 264005 China Jiangsu University Automotive Engineering Research Institute Zhenjiang 212013 China

Pedestrians’ intention to cross the street exercises a substantial influence on the decision-making process of autonomous vehicles in urban traffic environments. However, accurately predicting pedestrian crossing intention is non-trivial due to the interweaving of pedestrian personalities and traffic scene elements. Despite the significant achievement of previous studies, challenges remain in effectively extracting and integrating diverse features from different modalities of observation data. In response, this paper proposes a novel model leveraging pedestrian bounding boxes, poses, and ego-vehicle speed to predict crossing intention. We introduce mixture expert feature embedding (MEFE) to project raw data into high-dimensional space based on different types of inputs. A multi-branch spatial and temporal graph convolutional network (MB-STGCN) is applied to capture multi-scale spatial and temporal features of pedestrian pose skeleton joints. A multi-token temporal aggregation (MTTA) method is devised to preserve abundant temporal information of observation data. Additionally, a progressive multimodal feature fusion (PMFF) method based on symmetric channel split attention (SCSA) is employed to enhance the interaction between different modalities when integrating different features. Extensive comparison experiments and ablation studies on public benchmark datasets substantiate the effectiveness of our approach, showing significant improvements over existing models. © 2000-2011 IEEE.

关键词： attention mechanism Crossing intention prediction graph convolution multimodal feature fusion

来源：评论

学校读者我要写书评

暂无评论

Data Heterogeneity Challenge in Consumer Electronics: A Federated Learning Solution

引用

IEEE Transactions on Consumer Electronics 2025年

作者： Liu, Zhenyu Wu, Yunting Ge, Huanhuan Wang, Jinlong Lyu, Zhihan Qingdao University of Technology School of Information and Control Engineering Shandong Qingdao China School of Information and Control Engineering Innovation Institute for Sustainable Maritime Architecture Research and Technology Shandong Qingdao China School of Computer Science and Technology Xidian University Xi’an 710126 China

Federated learning(FL) enables collaborative model training across consumer electronic devices while preserving data privacy. However, the non-independent and identically distributed nature of data in real-world scenarios significantly impacts personalization performance. To address this challenge, we propose pFedPKD, in this framework, we first employ model pruning to simplify the global model, thereby enhancing its adaptability and improving the efficiency of knowledge transfer between different electronics devices. Moreover, we introduce an attention-based knowledge transfer mechanism, which leverages knowledge distillation to transfer the attention map and predictions of the pruned global model to the local models, thus improving their generalization capability and performance. Extensive experiments on three datasets demonstrate that pFedPKD consistently outperforms state-of-the-art methods, achieving higher personalization accuracy and robustness under heterogeneous data distributions. This work provides an efficient solution for applying FL in consumer electronics, offering improved adaptability and resource efficiency. © 1975-2011 IEEE.

关键词： consumer electronics Federated learning knowledge distillation Non-IID

来源：评论

学校读者我要写书评

暂无评论

Development of Fast Prediction System of User Action for a Fully Autonomous Smart Home System

Development of Fast Prediction System of User Action for a F...

引用

Sentiment Analysis and Deep Learning (ICSADL), International Conference on

作者： N.R. Rajalakshmi L. Sharmila S. Prabu T. Veeramakali G. Arulkumaran Pranshu Chourasia Department of Computer Science and Engineering Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and Technology Tamil Nadu India Department of Computing Technologies School of Computing SRM Institute of Science and Technology Kattankulathur Chennai India Department of Data Science and Business Systems School of Computing SRM Institute of Science and Technology Kattankulathur Chennai India School of Computing and Information Technology REVA UNIVERSITY Bangalore Karnataka India

ISBN: (数字)9798331523923

ISBN: (纸本)9798331523930

The HVAC system, energy storage building, distributed power supply, and other equipment are integrated into the scheduling algorithm, which is aimed at reducing household electricity consumption. It is also assumed that users can provide energy to the grid according to their own conditions. Taking electricity cost and comfort level as optimization targets, a home energy optimization control model for the coordinated management of hybrid energy sources is built. A smart scheduling mechanism based on the improved adaptive particle swarm optimization approach is proposed in order to derive the best time intervals for electric appliances, necessary power for the control of the room temperature for every time frame, and power for charging and discharging of the storage battery at various moments. Simulation results show that through the incorporation of distributed photovoltaic power generation, backup storage by battery, and home energy optimization control, the system efficiently balances between user comfort and electricity consumption. This offers great technical support to the development of home energy management systems. By using time-of-use electricity price for energy acquisition and supply, the optimization control goal is minimizing both power use and cost as well as preserving comfort levels. The hybrid energy management's proposed home energy optimization control model uses an adaptive particle swarm optimization algorithm to find the optimal operation schedules of the electrical appliances, the required power for temperature control in a room, and the charge/discharge power level of the storage battery at each time interval. As per the optimization principle, the proposed dynamic programming algorithm converts the multi-stage problem into a sequence of single-stage problems and solves them separately. This method successfully resolves intricate problems that cannot be addressed through greedy algorithms or divide-and-conquer. In this research, management ac

关键词： Management accounting Heuristic algorithms Electricity Prediction algorithms Dynamic programming Batteries Temperature control Particle swarm optimization Energy management systems Optimization

来源：评论

学校读者我要写书评

暂无评论

A Novel Tactile Sensing Skin for Surface Perception

引用

IEEE Sensors Journal 2025年

作者： Das, Subham Thiyagarajan, Karthick Kodagoda, Sarath Krishnan, Athul Bhattacharjee, Mitradip Indian Institute of Science Education and Research Bhopal i-Labs Electrical Engineering and Computer Science Department Madhya Pradesh 462066 India Centre for Advanced Manufacturing Technology School of Engineering Design and Built Environment PenrithNSW2751 Australia UTS Robotics Institute University of Technology Sydney Australia SydneyNSW2007 Australia

Humans employ a fusion of pressure and temperature signals to perceive tactile stimuli. While replicating the complexity of human tactile sensing poses challenges, advancements in artificial tactile sensing skins play a crucial role in the progress of robotics and prosthetics. Here, we present a novel tactile sensing skin designed for the perception of diverse surfaces. It utilizes a microstructured porous tactile sensor, exhibiting notable sensitivity at 21.5 kPa-1 under low-pressure conditions and 2.5 kPa-1 in high-pressure scenarios. The sensor also boasts an impressively low detection limit of 2.5 Pa, allowing for discernment of subtle pressure variations. Simultaneously, we utilize a flexible temperature sensor exhibiting a sensitivity of 0.58 % ∘C-1 and swift response attributes to facilitate the detection of radiated heat from diverse surfaces. The incorporation of a multilayer perceptron algorithm into the developed glove-like system, equipped with tactile and temperature sensors, achieves 87% accuracy in classifying eleven distinct surfaces. © 2001-2012 IEEE.

关键词： Tactile sensors

来源：评论

学校读者我要写书评

暂无评论

3-D RSSD Localization under Mixed Gaussian Noise and NLOS Environments in UWSNs

引用

IEEE Internet of Things Journal 2025年

作者： Zhang, Yuanyuan Gulliver, T. Aaron Wu, Huafeng Li, Jiping Mei, Xiaojun Xian, Jiangfeng Li, Kuan-Ching School of Computer and Information Engineering Changzhou213032 China Shanghai Maritime University China and the Merchant Marine College Shanghai201306 China University of Victoria Department of Electrical and Computer Engineering VictoriaBCV8W 2Y2 Canada Shanghai Maritime University Merchant Marine College Shanghai201306 China Changzhou Institute of Technology School of Computer and Information Engineering Changzhou213032 China Shanghai Maritime University Institute of Logistics Science and Engineering Shanghai201306 China Providence University Department of Computer Science and Information Engineering Taichung43301 Taiwan

This article presents a robust 3-D Received Signal Strength Difference (RSSD) localization algorithm under mixed Gaussian noise in Underwater Wireless Sensor Networks (UWSNs) with Non-Line-Of-Sight (NLOS) paths. To mitigate the adverse effects, concurrent to absorption and path losses on accurate underwater localization, an Efficient RSSD-based Iterative Estimator (ERIE) in mixed Gaussian noise and NLOS environments is proposed. First, the corresponding non-convex problem in such environments is formulated, and the direct solution to this problem is not tractable unfortunately. Considering underwater acoustic signal attenuation, an RSSD-based min-max strategy is designed to transform it into a problem minimizing the worst-case loss, combined with the Huber cost function, constitutes a Huber function-based equivalent problem (H-ADMM) solved by Alternating Direction Method of Multipliers (ADMM). A compensation matrix is designed based on the H-ADMM solution to compensate for the bias introduced by the transformation, and the corresponding Cramér-Rao Lower Bound (CRLB) is derived to provide a performance benchmark. Numerical results indicate that the proposed approach achieves a higher localization accuracy than state-of-the-art methods. © 2014 IEEE.

关键词： Cost functions

来源：评论

学校读者我要写书评

暂无评论

BeliN: A Novel Corpus for Bengali Religious News Headline Generation using Contextual Feature Fusion

arXiv

引用

arXiv 2025年

作者： Osama, Md Dey, Ashim Ahmed, Kawsar Kabir, Muhammad Ashad Department of Computer Science and Engineering Chittagong University of Engineering and Technology Chittagong Raozan4349 Bangladesh Saidpur Nilphamari5310 Bangladesh School of Computing Mathematics and Engineering Charles Sturt University BathurstNSW2795 Australia

Automatic text summarization, particularly headline generation, remains a critical yet underexplored area for Bengali religious news. Existing approaches to headline generation typically rely solely on the article content, overlooking crucial contextual features such as sentiment, category and aspect. This limitation significantly hinders their effectiveness and overall performance. This study addresses this limitation by introducing a novel corpus, BeliN (Bengali Religious News) - comprising religious news articles from prominent Bangladeshi online newspapers, and MultiGen - a contextual multi-input feature fusion headline generation approach. Leveraging transformer-based pre-trained language models such as BanglaT5, mBART, mT5, and mT0, MultiGen integrates additional contextual features-including category, aspect, and sentiment-with the news content. This fusion enables the model to capture critical contextual information often overlooked by traditional methods. Experimental results demonstrate the superiority of MultiGen over the baseline approach that uses only news content, achieving a BLEU score of 18.61 and ROUGE-L score of 24.19, compared to baseline approach scores of 16.08 and 23.08, respectively. These findings underscore the importance of incorporating contextual features in headline generation for low-resource languages. By bridging linguistic and cultural gaps, this research advances natural language processing for Bengali and other underrepresented languages. To promote reproducibility and further exploration, the dataset and implementation code are publicly accessible at https://***/akabircs/BeliN. © 2025, CC BY-NC-ND.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Cluster Head Selection Algorithm using Machine Learning

Cluster Head Selection Algorithm using Machine Learning

引用

Sentiment Analysis and Deep Learning (ICSADL), International Conference on

作者： M.S. Nidhya E. Gurumoorthi M. Babu A.C. Charumathi Mohd Naved Balajee Maram School of Computer Applications Dayananda Sagar University Bangalore School of Engineering Malla Reddy University Dept of AI & DS Chennai Institute of Technology college Chennai Dept of Information Technology Sona College of technology Dept of Business Analytics Jaipuria Institute of Management Noida School of Computer Science and Artificial Intelligence SR University Ananthasagar Hasanparthy Warangal

ISBN: (数字)9798331523923

ISBN: (纸本)9798331523930

Wireless Sensor Networks (WSNs) find extensive applications in environmental monitoring, healthcare, and smart cities. Energy efficiency, however, continues to be a significant challenge with the limited lifetime of sensor node batteries. The conventional heuristic-based cluster head (CH) selection techniques tend not to adapt dynamically to network changes, resulting in inefficient energy utilization and decreased network lifetime. This research discusses Machine Learning (ML)-oriented CH selection approaches to improve energy efficiency, network lifetime, and data fusion. Using supervised and unsupervised learning methods, ML algorithms are able to learn optimal CHs dynamically based on residual energy, network structure, and data traffic. The comparative analysis show that ML-based CH selection enhances network stability by 95% and lowers energy consumption by 50% compared to traditional methods. The research points to the promise of ML in WSN performance optimization and opens the doors to intelligent, adaptive clustering technologies.

关键词： Wireless sensor networks Machine learning algorithms Smart cities Heuristic algorithms Supervised learning Clustering algorithms Magnetic heads Energy efficiency Stability analysis Unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Decentralized State Estimation and Opacity Verification Based on Partially Ordered Observation Sequences

arXiv

引用

arXiv 2025年

作者： Sun, Dajiang Hadjicostis, Christoforos N. Li, Zhiwu School of Electro-Mechanical Engineering Xidian University Xi’an710071 China Department of Electrical and Computer Engineering University of Cyprus Nicosia Cyprus Institute of Systems Engineering Macau University of Science and Technology Taipa999078 China

In this paper, we investigate state estimation and opacity verification problems within a decentralized observation architecture. Specifically, we consider a discrete event system whose behavior is recorded by a set of observation sites. These sites transmit the partially ordered sequences of observations that they record to a coordinator whenever a synchronization occurs. To properly analyze the system behavior from the coordinator’s viewpoint, we first introduce the notion of an All Sequence Structure (ASS), which concisely captures the state evolution of each system state upon different information provided by the observation sites. Based on the ASS, we then construct corresponding current-state and initial-state estimators for offline state estimation at the coordinator. When used to verify state-isolation properties under this decentralized architecture, the use of ASS demonstrates a significant reduction in complexity compared with existing approaches in the literature. In particular, we discuss how to verify initial-state opacity at the coordinator, as well as a novel opacity notion, namely current-state-at-synchronization opacity. Copyright © 2025, The Authors. All rights reserved.

关键词： Synchronization

来源：评论

学校读者我要写书评

暂无评论

Unleashing the Potentials of IoT with Focus on Energy and Path Loss for Internet of Medical Things

引用

Journal of Circuits, Systems and computers 2025年

作者： Hussain, Altaf Li, Shuaiyong Attar, Razaz Waheeb Al Dabel, Maryam M. Bashir, Ali Kashif Alhomoud, Ahmed Hussain, Tariq School of Computer Science and Technology Chongqing University of Posts and Telecommunications Chongqing400065 China Key Laboratory of Industrial Internet of Things and Networked Control Ministry of Education Chongqing University of Posts and Telecommunications Chongqing400065 China Management Department College of Business Administration Princess Nourah bint Abdulrahman University P.O. Box 84428 Riyadh11671 Saudi Arabia Department of Computer Science and Engineering College of Computer Science and Engineering University of Hafr Al Batin Saudi Arabia Department of Computing and Mathematics Manchester Metropolitan University United Kingdom Centre for Research Impact & Outcome Chitkara University Institute of Engineering and Technology Chitkara University Punjab Rajpura140401 India Department of Computer Science College of Science Northern Border University Arar91431 Saudi Arabia School of Computer Science and Technology Zhejiang Gongshang University Hangzhou310018 China

The Internet of Things (IoT) encompasses a broad platform of sensor networks incorporating independent wireless networks. With advancements in sensor technology and IoT-enabled networks, their applications in the medical field have led to the development of the Internet of Medical Things (IoMT). In IoMT, sensor nodes monitor and evaluate patient conditions such as heartbeat, blood sugar levels, blood pressure and temperature, and can also remotely track patient activities through remote analysis. These IoMT systems utilize tiny sensors with limited communication ranges to gather essential patient information. Wireless devices are equipped with a short range and need a direct communication path. However, transmitting data from the source node to the destination node ultimately results in energy consumption and path loss. Path loss models and energy consumption models are essential to address these issues. In this paper, we propose a novel routing protocol named Energy Efficient and Path Loss Preserving (EEPLP) for IoMT. The EEPLP protocol focuses on energy efficiency and path loss preservation based on the relay approach. Two models are being proposed, one for path loss and the other for energy consumption. Finally, both models are merged since the major contribution is to avoid path loss and enhance the protocol's energy efficiency. The EEPLP evaluates the state-of-the-art existing approaches of IoMT. The protocol is evaluated by simulating conditions and compared with other similar routing protocols already deployed in the IoMT;it has been observed that the EEPLP scheme has the potential to be maneuvered in IoMT structures with core targets of energy efficiency as well as path loss preservation techniques. © 2025 World Scientific Publishing Company.

关键词： Sensor nodes

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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：