检索结果-内蒙古大学图书馆

Personalized federated few-shot node classification

science China(Information sciences) 2025年第1期68卷 204-218页

作者： Yunfeng ZHAO Xintong HE Guoxian YU Jun WANG Yongqing ZHENG Carlotta DOMENICONI School of Software Shandong University Joint SDU-NTU Centre for Artificial Intelligence Research Shandong University Department of Mathematics National University of Singapore Department of Computer Science George Mason University

Personalized federated learning(PFL) aims to train customized models for individual clients in a decentralized setting, with the account of non-independent and identically distributed data across clients. However, most PFL methods adopt uniform classification layers for diverse clients and give rise to error-prone predictions, due to the task heterogeneity notably prominent in decentralized graph data scenarios. Although some PFL solutions setup client-specific classification layers for each client and optimize them only locally, they are corrupted with limited local training data. We propose an innovative solution called federated parameter decoupling and node augmentation(Fed PANo) to address these problems and to achieve personalized federated few-shot node classification, which is a prevalent and challenging but unexplored topic. Specifically, Fed PANo first separates the local model into the GNN and classifier to handle unique client-specific task variations. The GNN is trained through federated learning to capture shared knowledge of graph nodes across clients, while the classifier is custom-designed and trained individually for each client. Additionally, a generic classifier shared among clients is adopted to encourage the GNN's grasp of shared information. Then Fed PANo further proposes the node generator along with its local and collaborative training strategies to deal with the node scarcity of clients. Extensive experimental results on benchmark datasets confirm that Fed PANo outperforms eight competitive baselines across different settings.

关键词： personalized federated learning few-shot learning node classification task heterogeneity parameter decoupling node augmentation

来源：评论

学校读者我要写书评

暂无评论

Enhancing single-lead electrocardiogram arrhythmia detection with empirical mode decomposition

引用

Neural Computing and Applications 2025年 1-23页

作者： Issa, Mohamed F. Yousry, Ahmed Tuboly, Gergely Wang, Zeyu Zoltan, Juhasz Selim, Mazen M. AbuEl-Atta, Ahmed H. Department of Scientific Computing Faculty of Computers and Artificial Intelligence Benha University Benha13511 Egypt Department of Computer Science Faculty of Computers and Artificial Intelligence Benha University Benha13511 Egypt Department of Electrical Engineering and Information Systems University of Pannonia Veszprem8200 Hungary Department of Mechatronics Delta University for Science and Technology Gamasa35712 Egypt

Cardiac arrhythmias pose a significant challenge to health care, requiring accurate and reliable detection methods to enable early diagnosis and treatment. However, traditional ECG beat classification methods often lack robustness and fail to generalize effectively to diverse clinical scenarios, particularly when using single-lead-I recordings, which are less studied compared to lead II. To address this gap, this paper proposes a novel method that combines deep learning and empirical mode decomposition (EMD) for arrhythmia detection using single-lead-I ECG recordings obtained with electrodes on each hand. The hybrid model comprises 149 layers of one-dimensional convolution, max pooling, and residual blocks. By leveraging EMD, it classifies ECG beats into normal beats, premature ventricular contractions, and atrial premature beats with high sensitivity (99.86%, 99.72%, and 99.84%, respectively) using the first four intrinsic mode functions (IMFs). The proposed method is applicable in clinical and out-of-hospital monitoring scenarios, and an accompanying web application provides real-time analysis and diagnostic results. This approach addresses a critical gap in lead-I ECG analysis, enhancing diagnostic precision and expanding the scope of ECG applications in health care. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Empirical mode decomposition

来源：评论

学校读者我要写书评

暂无评论

Webly Supervised Fine-Grained Classification by Integrally Tackling Noises and Subtle Differences

引用

IEEE Transactions on Image Processing 2025年 34卷 2641-2653页

作者： Chen, Junjie Yan, Jiebin Fang, Yuming Niu, Li Jiangxi University of Finance and Economics School of Computing and Artificial Intelligence Jiangxi Nanchang330032 China Shanghai Jiao Tong University Department of Computer Science and Engineering Shanghai China

Webly-supervised fine-grained visual classification (WSL-FGVC) aims to learn similar sub-classes from cheap web images, which suffers from two major issues: label noises in web images and subtle differences among fine-grained classes. However, existing methods for WSL-FGVC only focus on suppressing noise at image-level, but neglect to mine cues at pixel-level to distinguish the subtle differences among fine-grained classes. In this paper, we propose a bag-level top-down attention framework, which could tackle label noises and mine subtle cues simultaneously and integrally. Specifically, our method first extracts high-level semantic information from a bag of images belonging to the same class, and then uses the bag-level information to mine discriminative regions in various scales of each image. Besides, we propose to derive attention weights from attention maps to weight the bag-level fusion for a robust supervision. We also propose an attention loss on self-bag attention and cross-bag attention to facilitate the learning of valid attention. Extensive experiments on four WSL-FGVC datasets, i.e., Web-Aircraft, Web-Bird, Web-Car, and WebiNat-5089, demonstrate the effectiveness of our method against the state-of-the-art methods. © 1992-2012 IEEE.

关键词： Labeled data

来源：评论

学校读者我要写书评

暂无评论

Groundwater Level Forecasting using Spatio-temporal Graph Convolutional Networks (GCNs) and XGBoost 6

Groundwater Level Forecasting using Spatio-temporal Graph Co...

引用

6th International Conference on Mobile Computing and Sustainable Informatics, ICMCSI 2025

作者： Naga Satish, G. Jadala, Vijaya Chandra Sravani, Kolisetti Giridhara Allam, Sangeetha Shanmuga Sundari, M. Thada, Jyothirmai BVRIT HYDERABAD College of Engineering for Women Hyderabad India Department of Computer Science and Artificial Intelligence School of Computer Science and Artificial Intelligence SR University Warangal506371 India Department of AIML Gokaraju Rangaraju Institute of Engineering And Technology Hyderabad India Department of Computer Science and Engineering MLR Institute of Technology Hyderabad India

ISBN: (纸本)9798331522667

This research analyzes groundwater levels across multiple districts using data from over 100 observation wells in each district. To capture seasonal variations and predict groundwater behavior, this research has developed three models: periodic, polynomial, and rainfall-based. The periodic and polynomial models describe groundwater level fluctuations based on historical well data, while the rainfall model assesses the influence of precipitation on water levels in the wells. In addition, this research explores advanced predictive techniques by incorporating Spatio-temporal Graph Convolutional Networks (GCNs) and XGBoost. These methods enable a more nuanced understanding of spatio-temporal dependencies and improve predictive accuracy by leveraging both spatial relationships among wells and the temporal evolution of groundwater levels. The integration of traditional models with machine learning techniques aims to enhance groundwater management and inform decision-making. © 2025 IEEE.

关键词： Rain

来源：评论

学校读者我要写书评

暂无评论

Real-Time Driver Alertness Assessment Using Neural Networks

Real-Time Driver Alertness Assessment Using Neural Networks

引用

2025 International Conference on Multi-Agent Systems for Collaborative intelligence, ICMSCI 2025

作者： Yoga, M. Ramyasri, M.M. Lokitha, R. Jane Karunya, J. Jeeva, N. Kokila, M.M. Kongu Engineering College Department of Artificial Intelligence Tamilnadu India Nandha Engineering College Department of Artificial Intelligence and Data Science Tamilnadu India

ISBN: (数字)9798331509828

ISBN: (纸本)9798331509828

Drivers have profited significantly from developments in computer technology with the introduction of intelligent car systems. However, driver weariness is a key contributing cause to many car accidents. This research suggests an intelligent system for detecting driver tiredness based on vision, employing a convolutional neural network (CNN) to capture latent facial traits rather than relying on hand-engineered features like blink rate, eye closure, and yawning. The driver is classified as either sleepy or not by a softmax layer, which enables the system to alert the driver and avoid collisions. The proposed approach relies on extracting features from driver photos using a CNN and assessing eye states and facial expressions to reliably identify tiredness. The algorithm's excellent accuracy in identifying sleepy states is demonstrated by the results, underscoring its potential to improve road safety and lower the number of accidents caused by fatigued drivers. © 2025 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

LesionAid: vision transformers-based skin lesion generation and classification – A practical review

引用

Multimedia Tools and Applications 2025年 1-22页

作者： K, Mallikharjuna Rao Krishna, Ghanta Sai Supriya, Kundrapu Sorgile, Meetiksha Data Science and Artificial Intelligence International Institute of Information Technology Naya Raipur India Computer Science and Engineering International Institute of Information Technology Naya Raipur India

Skin cancer is one of the most prevalent forms of human cancer. It is recognized mainly visually, beginning with clinical screening and continuing with the dermoscopic examination, histological assessment, and specimen collection. Deep convolutional neural networks (CNNs) perform highly segregated and potentially universal tasks against a classified fine-grained object. This study suggests a novel multi-class prediction framework that uses ViT and ViTGAN to categories skin lesions. To address the class disparity, GANs (Generative Adversarial Networks) based on vision transformers are used. The framework comprises four main phases: ViTGANs, Image processing, and explainable AI. Phase 1 consists of generating synthetic images using ViTGAN to balance all the classes in the dataset. To enhance the amount of the data, the second phase involves using various morphological processes and data augmentation techniques. In phases three and four, a ViT model for edge computing systems that can recognize patterns and classify skin lesions from the user's skin that is visible in the picture is developed. In phase 3, after classifying the lesions into the desired class with ViT, we will use explainable AI (XAI) that leads to more explainable results (using activation maps, etc.) while ensuring high predictive accuracy. The results demonstrate that the model used for generating synthetic images has achieved an FID score of 13.32 and the ViT model has achieved 99.2% as its training accuracy and 97.4% as its validation accuracy. The whole framework is compared with the existing frameworks for skin lesion detection. And explainable AI (XAI) is used in the proposed framework in order to increase model transparency and boost trust among users by illustrating the main factors affecting predictions. This interpretability helps physicians in making more informed decisions by providing clear insights into the model's reasoning. © The Author(s), under exclusive licence to Springer science+Bu

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

A Fuzzy Multi-Objective Framework for Energy Optimization and Reliable Routing in Wireless Sensor Networks via Particle Swarm Optimization

引用

computers, Materials & Continua 2025年第5期83卷 2773-2792页

作者： Medhat A.Tawfeek Ibrahim Alrashdi Madallah Alruwaili Fatma M.Talaat Department of Computer Science College of Computer and Information SciencesJouf UniversitySakaka72341AljoufSaudi Arabia Department of Computer Engineering and Networks College of Computer and Information SciencesJouf UniversitySakaka72341AljoufSaudi Arabia Faculty of Artificial Intelligence Kafrelsheikh UniversityKafrelsheikh33516Egypt Faculty of Computer Science&Engineering New Mansoura UniversityGamasa35712Egypt

Wireless Sensor Networks(WSNs)are one of the best technologies of the 21st century and have seen tremendous growth over the past *** work has been put into its development in various aspects such as architectural attention,routing protocols,location exploration,time exploration,*** research aims to optimize routing protocols and address the challenges arising from conflicting objectives in WSN environments,such as balancing energy consumption,ensuring routing reliability,distributing network load,and selecting the shortest *** optimization techniques have shown success in achieving one or two objectives but struggle to achieve the right balance between multiple conflicting *** address this gap,this paper proposes an innovative approach that integrates Particle Swarm Optimization(PSO)with a fuzzy multi-objective *** proposed method uses fuzzy logic to effectively control multiple competing objectives to represent its major development beyond existing methods that only deal with one or two *** search efficiency is improved by particle swarm optimization(PSO)which overcomes the large computational requirements that serve as a major drawback of existing *** PSO algorithm is adapted for WSNs to optimize routing paths based on fuzzy multi-objective *** fuzzy logic framework uses predefined membership functions and rule-based reasoning to adjust routing *** adjustments influence PSO’s velocity updates,ensuring continuous adaptation under varying network *** proposed multi-objective PSO-fuzzy model is evaluated using NS-3 *** results show that the proposed model is capable of improving the network lifetime by 15.2%–22.4%,increasing the stabilization time by 18.7%–25.5%,and increasing the residual energy by 8.9%–16.2% compared to the state-of-the-art *** proposed model also achieves a 15%–24% reduction in load variance,demonstrating balanced routing and extended net

关键词： Wireless sensor networks particle swarm optimization fuzzy multi-objective framework routing stability

来源：评论

学校读者我要写书评

暂无评论

A novel approach for image retrieval in remote sensing using vision-language-based image caption generation

引用

Multimedia Tools and Applications 2025年第6期84卷 2985-3014页

作者： Yadav, Prem Shanker Tyagi, Dinesh Kumar Vipparthi, Santosh Kumar Department of Computer Science and Engineering Malaviya National Institute of Technology Rajasthan Jaipur302017 India School of Artificial Intelligence and Data Engineering Indian Institute of Technology Ropar Punjab Rupnagar140001 India

Recent advancements in satellite technologies have resulted in the emergence of Remote Sensing (RS) images. Hence, the primary imperative research domain is designing a precise retrieval model for retrieving the most pertinent images based on the query. Present Remote Sensing Image Retrieval (RSIR) systems use visual descriptors to characterize the primitives (such as various land-cover types) that are visible in the images. However, the visual descriptors are inadequate for defining the complicated content of RS images. To solve this problem, a new model is devised for image retrieval based on image captions. The goal is to generate textual illustrations with captions to define relations amongst objects precisely. Here, image captioning is attained based on the vision-language pre-training model. The image captions are utilized for generating features like term frequency-inverse document frequency (TF-IDF), length of text, and Bag of Words. Meanwhile, query text is utilized wherein features like TF-IDF, text length, and Bag of Words are obtained. The similarity between query text features and the image captions features has been computed on the basis of a hybrid similarity measure wherein weights are tuned with the proposed Honey Badger Political Optimizer (HBPO) to retrieve the image. The proposed HBPO provided enhanced efficiency with elevated precision of 93.3%, recall of 93.7%, F1-score of 93.5%, and Recall-Oriented Understudy for Gisting Evaluation (ROUGE) of 0.441. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

Leveraging BERT, DistilBERT and TinyBERT for Rumor Detection

引用

IEEE Access 2025年 13卷 72918-72929页

作者： Qazi, Aijazahamed Goudar, R.H. Patil, Rudragoud Hukkeri, Geetabai S. Kulkarni, Dhanashree Angadi Institute of Technology and Management Department of Artificial Intelligence and Data science Belagavi591108 India Visvesvaraya Technological University Department of Computer Science and Engineering Belagavi590018 India KLS Gogte Institute of Technology Department of Computer Science and Engineering Belagavi590008 India Manipal Academy of Higher Education Manipal Institute of Technology Department of Computer Science and Engineering Bengaluru Karnataka Manipal576104 India Angadi Institute of Technology and Management Department of Computer Science and Engineering Belagavi591108 India

The rapid spread of false information on social media has become a major challenge in today's digital world. This has created a need for an effective rumor detection system that can identify and control the spread of false information in real-time. The proposed work introduces a rumor detection system by integrating transformer-based models such as BERT, DistilBERT, and TinyBERT with traditional Machine Learning (ML) techniques. The proposed method involves fine-tuning transformer models to recognize the contextual meaning of social media posts. The classifiers include Decision Tree (DT), Support Vector Machine (SVM), Random Forest (RF) and Naïve Bayes (NB), which help in categorizing content as either rumor or non-rumor based on the patterns. The proposed work evaluated BERT, DistilBERT, TinyBERT combined with ML models (SVM, DT, RF, NB) across PHEME dataset using 70:30, 60:40, and 80:20 splits. Overall, BERT+DT and TinyBERT+SVM provided significant results, with BERT+RF and DistilBERT+NB demonstrating better classification capabilities across various events and split ratios on the dataset. © 2013 IEEE.

关键词： Tweets

来源：评论

学校读者我要写书评

暂无评论

Deep Learning for Localization of White Matter Lesions in Neurological Diseases 6

Deep Learning for Localization of White Matter Lesions in Ne...

引用

6th Northern Lights Deep Learning Conference, NLDL 2025

作者： Machnio, Julia Nielsen, Mads Ghazi, Mostafa Mehdipour Pioneer Centre for Artificial Intelligence Department of Computer Science University of Copenhagen Denmark

White Matter (WM) lesions, commonly observed as hyperintensities on FLAIR MRIs or hypointensities on T1-weighted images, are associated with neurological diseases. The spatial distribution of these lesions is linked to an increased risk of developing neurological conditions, emphasizing the need for location-based analyses. Traditional manual identification and localization of WM lesions are labor-intensive and time-consuming, highlighting the need for automated solutions. In this study, we propose novel deep learning-based methods for automated WM lesion segmentation and localization. Our approach utilizes state-of-the-art models to concurrently segment WM lesions and anatomical WM regions, providing detailed insights into their distribution within the brain’s anatomical structure. By applying k-means clustering to the regional WM lesion load, distinct subject groups are identified to be associated with various neurological conditions, validating the method’s alignment with established clinical findings. The robustness and adaptability of our method across different scanner types and imaging protocols make it a valuable tool for research and clinical practice, offering potential improvements in diagnostic efficiency and patient care. Codes and refined atlas utilized in this study are available at https://***/juliamachnio/WMHLocalization. © NLDL *** rights reserved.

关键词： Risk assessment

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：