检索结果-内蒙古大学图书馆

20th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2024

作者： An, Ledong Wang, Chenyang Huang, Shaoyuan Guo, Yang Zhang, Cheng Qiu, Chao Wang, Xiaofei College of Intelligent and Computing Tianjin University Tianjin China College of Computer Science and Software Engineering Shenzhen University Guangdong Shenzhen518060 China Tianjin University of Finance and Economics Tianjin China Software Technology Company Limited Suzhou215163 China

ISBN: (纸本)9789819628292

The rapid advancements in time-series forecasting have led to the emergence of datasets with increasingly diverse characteristics. Researchers typically focus on designing robust algorithms to handle these datasets. However, model performance can vary across different datasets. Existing studies commonly use Mean Absolute Error (MAE) and Mean Squared Error (MSE) to evaluate model performance. These metrics often need to fully account for the impact of dataset quality on forecast accuracy and reliability, leading to insufficient explanations of model forecasts and even dataset dependence. Furthermore, the lack of dataset evaluation makes it difficult to determine whether forecast results are influenced by data characteristics or model architecture, significantly hindering model improvement. To address this problem, this study proposes a new dataset evaluation paradigm-Multiscale Temporal Entropy Evaluation Paradigm. This paradigm aims to tackle the problem of data feature ambiguity affecting model forecast interpretability. By evaluating nine time-series datasets from domains such as weather, economics, transportation, and networks and comparing these evaluations with the forecast results from six different models, this study demonstrates the effectiveness of this method in explaining the relationship between dataset features and model performance. © IFIP International Federation for Information Processing 2025.

关键词： Spatio-temporal data

来源：评论

学校读者我要写书评

暂无评论

MULTI-TARGET VITAL SIGN DETECTION BY FUSION OF BIOLOGICAL RADAR AND CONVOLUTIONAL NEURAL NETWORK

引用

Scalable Computing 2025年第3期26卷 1496-1503页

作者： Yuan, Hongbin Yuan†, Chenyao Cao, Huiqun School of Telecommunications and Intelligent Manufacturing Sias University Henan Zhengzhou451150 China School of Computer and Software Engineering Sias University Zhengzhou451150 China Academic Affairs Office of Kaifeng Modern Technology Secondary Vocational School Henan Kaifeng475000 China

In order to address the increasing demand for vital sign detection, the author proposes a multi-target vital sign detection research that combines biological radar and convolutional neural network. Based on the fundamental architecture of convolutional neural networks (CNNs), the author combines classification-based CNN object detection techniques to develop a biological radar multi-target vital sign detection platform. The feasibility of this approach is confirmed through experiments, demonstrating the integration of biological radar and CNNs for multi-target vital sign detection. The experimental results indicate that the biological radar achieves a recognition accuracy of 96.1%, proving the effectiveness of the biological radar detection algorithm. The research on multi-target vital sign detection based on the fusion of biological radar and convolutional neural network is an effective auxiliary method that can provide reference for relevant researchers. © 2025 SCPE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Joint-Wise Distributed Perception Graph Convolutional Network for Skeleton-Based Action Recognition

Joint-Wise Distributed Perception Graph Convolutional Networ...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Huang, Qian Geng, Qiang Chen, Zhaoyu Li, Xin Li, Yangyang Li, Xing College of Computer Science and Software Engineering Hohai University China College of Information Science and Technology & Artificial Intelligence Nanjing Forestry University China

ISBN: (纸本)9798350368741

Recent studies have achieved remarkable results for action recognition with human skeletal data by utilizing graph convolutional models. Traditional approaches typically aggregate local spatio-temporal information bottom-up to form a single spatio-temporal global understanding. However, this method may fail to achieve multi-location perception for fine-grained action capture and may struggle to model subactions due to their varying durations. To solve these challenges, we design a Distributed Spatio-temporal Perception (DSP) module that innovatively treats each joint as an independent perception unit, performing joint-wise distributed multi-location perception in both spatial and temporal dimensions. In addition, we introduce an Anchor Pose-driven Subaction Encoding (APSE) module, which enhances informative clues for subaction reasoning through identifying the correlations between anchor pose and subactions formed by integrating distributed spatio-temporal perception features. Based on the above work, we propose a Joint-wise Distributed Perception Graph Convolutional Network (JDP-GCN). The experiments on three widely used datasets: NTU RGB+D 60, NTU RGB+D 120, and NW-UCLA, demonstrate that our method achieves state-of-the-art performance. © 2025 IEEE.

关键词： Distributed Perception Graph Convolutional Network Human Action Recognition Skeleton Modality

来源：评论

学校读者我要写书评

暂无评论

Detecting Cross-Domain Ambiguity In Requirements Through Natural Language Processing, A Systematic Literature Review

Detecting Cross-Domain Ambiguity In Requirements Through Nat...

引用

International Conference on Advancements in Computational Sciences (ICACS)

作者： Ibrahim Khalil Israr Ahmad Uzair Rasheed Wasi Haider Butt Zaeem Anwaar Department of Computer and Software Engineering CEME NUST Islamabad Pakistan

ISBN: (数字)9798331533038

ISBN: (纸本)9798331533045

A strong foundation for a project is the first step to measuring or estimating the success parameters of a project. Requirement engineers gather and elicit the requirements of the project in detail after the feasibility study. A project’s success can be measured/estimated if and only if the initially collected requirements are clear, unambiguous, and well-understood. Similarly, ambiguous or not understandable requirements can lead to failure or closure of the project in disastrous form. An initial step in requirement elicitation is usually gathering requirements in natural language. This study analyzes different tools, techniques, and approaches in different research articles used for detecting ambiguities in requirements in the natural language. We identified 33 research articles, published during 2012-24 through a Systematic Literature Review. For automated ambiguities detection in requirements engineering, we identified 17 tools & techniques and 12 NLP approaches. We analyzed the better approach for cross-domain ambiguity detection. SGNS (Skipgram Negative Sampling) variant of Word2Vec for the cross-domain ambiguity detection in the review of the study, resulted to be the most efficient technique implemented with the usage view and easily scalable for multiple requirements and domains.

关键词： Scalability software algorithms Natural language processing software Requirements engineering Systematic literature review

来源：评论

学校读者我要写书评

暂无评论

Dynamic Synergy: Leveraging RIS and Reinforcement Learning for Secure, Adaptive Underlay Cognitive Radio Networks

Dynamic Synergy: Leveraging RIS and Reinforcement Learning f...

引用

International Global Information Infrastructure Symposium (GIIS)

作者： Deemah H. Tashman Soumaya Cherkaoui Department of Computer and Software Engineering Polytechnique Montreal Montreal Canada

ISBN: (数字)9798331520113

ISBN: (纸本)9798331520120

Reconfigurable intelligent surfaces (RISs) have recently been employed to facilitate communication and improve performance by reflecting signals through configuring phase shifts toward the intended destination. This article examines the physical layer security of an underlay cognitive radio network aided by an RIS and in the presence of multiple eavesdroppers. The research is conducted under practical conditions, encompassing RIS hardware constraints and cascaded fading channels. An optimization problem is proposed with the objective of maximizing the secrecy rate of secondary users by optimizing the reflection angles of the RIS and the transmission power of the secondary user transmitter. A deep reinforcement learning method, specifically the soft actor-critic, is presented as a solution. The results section demonstrates the effect of altering the number of RIS elements on security. We also analyze the impact of hardware limitations and cascade levels on the secrecy rate. The effect of varying the number of eavesdroppers and the maximum permissible transmission power are also examined.

关键词： Fading channels Radio transmitters Reconfigurable intelligent surfaces Physical layer security Deep reinforcement learning Hardware Reflection Cognitive radio Security Optimization

来源：评论

学校读者我要写书评

暂无评论

Content Storage Management and Precaching Scheme in Content-Centric-Network-Based Internet of Vehicles

引用

IEEE Internet of Things Journal 2025年第9期12卷 12927-12947页

作者： Nam, Youngju Choi, Hyeonseok Lee, Euisin Kunsan National University School of Software Gunsan54150 Korea Republic of Research Institute for Computer and Information Communication Chungbuk National University Cheongju28644 Korea Republic of Chungbuk National University School of Information and Communication Engineering Cheongju28644 Korea Republic of

The emergence of smart cars has led to a significant increase in mobile data traffic on the backhaul links as a vast amount of contents is generated and consumed in the Internet of Vehicles (IoV). To cope with this situation, the precaching research of content-centric networks (CCNs) has been applied as a promising solution for reducing traffic consumption. However, most precaching schemes in CCN-IoV only focus on delay-sensitive content (DSC) and do not have proper content storage management (CSM), limiting their ability to minimize delays. To address these issues, we thus propose a novel CSM and precaching (CSMP) scheme consisting of three methods. To prevent the erases of precached or popular content, we have designed a CSM method based on our caching priority algorithm. Additionally, we have designed a DSC precaching method to improve mobility prediction and reduce delays using the recalculation time based on the Gaussian distribution with skewness. Finally, we have designed a delay-tolerant content (DTC) precaching method to minimize backhaul link traffic consumption by using an integer linear programming approach. The proposed CSMP scheme improved the evaluation value considering the success ratio and the backhaul link traffic by 23.21% compared with the existing schemes in our NS3-based simulations. © 2014 IEEE.

关键词： Delay tolerant networks

来源：评论

学校读者我要写书评

暂无评论

DocTTT: Test-Time Training for Handwritten Document Recognition Using Meta-Auxiliary Learning

arXiv

引用

arXiv 2025年

作者： Gu, Wenhao Gu, Li Wang, Ziqiang Suen, Ching Yee Wang, Yang Department of Computer Science and Software Engineering Concordia University Canada

Despite recent significant advancements in Handwritten Document Recognition (HDR), the efficient and accurate recognition of text against complex backgrounds, diverse handwriting styles, and varying document layouts remains a practical challenge. Moreover, this issue is seldom addressed in academic research, particularly in scenarios with minimal annotated data available. In this paper, we introduce the DocTTT framework to address these challenges. The key innovation of our approach is that it uses test-time training to adapt the model to each specific input during testing. We propose a novel Meta-Auxiliary learning approach that combines Meta-learning and self-supervised Masked Autoencoder (MAE). During testing, we adapt the visual representation parameters using a self-supervised MAE loss. During training, we learn the model parameters using a meta-learning framework, so that the model parameters are learned to adapt to a new input effectively. Experimental results show that our proposed method significantly outperforms existing state-of-the-art approaches on benchmark datasets. Copyright © 2025, The Authors. All rights reserved.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

DIG: Improved DINO for Graffiti Detection

引用

IEEE Transactions on Systems, Man, and Cybernetics: Systems 2025年第5期55卷 3557-3569页

作者： Wang, Bingshu Mao, Qianchen Liu, Aifei Chen, Long Chen, C. L. Philip Northwestern Polytechnical University School of Software Xi’an710129 China Shenzhen University Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Key Laboratory of Media Security Shenzhen518060 China University of Macao Department of Computer and Information Science China South China University of Technology School of Computer Science and Engineering Guangzhou510641 China

Graffiti detection is essential in historic building protection and urban neighborhood management. Graffiti detection has made significant progress in recent years based on the development of deep learning. However, small-scale graffiti, interference from the background, and the false detection of word parts in graffiti make it a challenging problem. This article proposes a Transformer-based high-precision graffiti detection method, namely DIG. Precisely, it consists of three modules: 1) Spatial query selection (SQS), scale-aware IoU loss (SIL);2) Denoising Task with binary contrastive denoising (BCDN);and 3) IoU-guided box denoising (IBD) modules. To detect small-scale graffiti, this paper proposes SIL to help the loss function to perceive small-scale graffiti and large-scale graffiti fairly. To reduce the false detection of word parts, this article presents the SQS module, which integrates spatial information into the query selection process of the Encoder to filter out falsely detected bounding boxes within the graffiti. To reduce the interference from the background, this article introduces a denoising task with BCDN and IBD modules, improving the model’s ability to distinguish graffiti from the background and accurately select appropriate bounding boxes. A large number of experimental results on the STORM dataset show that our method achieves state-of-the-art results with an AP50 of 87.9%. Moreover, DIG achieved competitive results on the FineFM dataset for mask detection. This indicates that DIG can also be conveniently transferred to detect other scenarios. © 2013 IEEE.

关键词： Historic preservation

来源：评论

学校读者我要写书评

暂无评论

A Comparison of Deep Learning and Machine Learning Methods for Water Quality Prediction 3

A Comparison of Deep Learning and Machine Learning Methods f...

引用

3rd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2025

作者： Srinivasa Gupta, N. Valarmathi, B. Hiwanj, Ishaan Department of Manufacturing Engineering School of Mechanical Engineering Vellore Institute of Technology Tamil Nadu Vellore India Department of Software and Systems Engineering School of Computer Science Engineering and Information Systems Vellore Institute of Technology Tamil Nadu Vellore India Department of Information Technology School of Computer Science Engineering and Information Systems Vellore Institute of Technology Tamil Nadu Vellore India

ISBN: (纸本)9798331527549

Predicting water quality is essential to preserving human health and environmental sustainability. Traditional water quality assessment methods often face scalability and real-time monitoring limitations. With accuracies of 62%, 72%, 83%, 69%, 63%, 66%, 71%, 63%, and 64%, respectively, the current techniques utilized were Logistic Regression, Decision Trees, Random Forest Regressor, Extreme Gradient Boosting, Naive Bayes, K-nearest neighbors, Support Vector Machine, AdaBoost, and Bagging [9]. This study addresses these challenges by leveraging Adaptive Synthetic Sampling (ADASYN) to balance the dataset and evaluating model performance on datasets of 5,000 and 10,000 entries per class. A robust dataset obtained from Kaggle was used, with five models – Long Short-Term Memory (LSTM), Feed Forward Neural Network (FFNN), Categorical Boosting (CatBoost), Extreme Gradient Boosting (XGBoost), and Random Forest – evaluated and compared. The proposed methods demonstrate significant improvements in accuracy, with XGBoost achieving the highest accuracy of 95.53%, followed by Random Forest at 93.98%. This work underscores the importance of advanced machine learning techniques in addressing the limitations of traditional methods, enhancing accuracy, scalability, and adaptability in water quality prediction. These findings contribute to advancing environmental monitoring and management practices with reliable, data-driven insights. © 2025 IEEE.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Clustering-Based Intrusion Detection System Meets Multi-Critics Generative Adversarial Networks

引用

IEEE Internet of Things Journal 2025年

作者： Wang, Haofan Kandah, Farah Mendis, Thilina Medury, Lalith Auburn University Computer Science and Software Engineering AuburnAL36849 United States

Network security has continuously been a major focus of research and concern on a global scale. The Intrusion Detection System (IDS), as a crucial defensive measure against network attacks, has undergone multiple iterations and evolutions since its inception to adapt to the ever-changing network environment. Due to the widespread issue of data imbalance in network security datasets, a single machine learning or deep learning model often struggles to effectively handle different types of attacks. In this work, we propose a Multi-Critics Generative Adversarial Networks (GAN) Clustering-Based IDS (MCGC-IDS) model to address the issue of data imbalance. The quality of the generated data is analyzed using correlation heatmaps and PCA plots, which later is used to update the dataset that is utilized for feature extraction with autoencoders. Subsequently, CNN-LSTM models are employed to analyze clusters formed by the Weighted Fuzzy C-means (WFCM) clustering algorithm to achieve enhanced performance for the IDS system. This model is then compared with two existing models. The results indicate that while the GAN-generated data retains the original dataset distribution, it also addresses the issue of imbalance. Moreover, the subsequent multi-layered processing enables the overall model to more effectively handle various types of attacks. Finally, when this model is tested on a similar dataset, the UNSW-NB15, it continues to demonstrate superior performance, indicating its strong generalizability. © 2014 IEEE.

关键词： Generative adversarial networks

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：