检索结果-内蒙古大学图书馆

19th International Conference on Ubiquitous information Management and Communication, IMCOM 2025

作者： Mir, Aabid A. Zuhairi, Megat F. Musa, Shahrulniza Namoun, Addallah Malaysian Institute of Information Technology Universiti Kuala Lumpur Kuala Lumpur Malaysia AI Centre Faculty of Computer and Information Systems Islamic University of Madinah Madinah Saudi Arabia

ISBN: (纸本)9798331507817

The inability to capture the temporal dynamics of network interactions limits traditional intrusion detection systems (IDSs) in detecting sophisticated threats that evolve over time. This research introduces DynKDD, a dynamic graph-based version of the NSL-KDD dataset, designed to capture the temporal relationships between network entities and their interactions. The proposed solution leverages Graph Neural Networks (GNNs) to detect network intrusions by analysing both the spatial and temporal evolution of network traffic. The methodology involves transforming the static NSL-KDD dataset into a series of time-ordered graph snapshots by assigning synthetic timestamps and constructing dynamic graphs. The snapshots represent the evolution of network interactions, enabling advanced dynamic graph learning models to process and detect intrusions. Five state-of-the-art GNN models, Graph Convolutional Networks (GCN), Graph Attention Networks (GAT), Graph Sample and Aggregation (GraphSAGE), Graph Isomorphism Networks (GIN) and Graph Auto-Encoders (GAE), were trained and tested on the DynKDD dataset, and their performance was evaluated using metrics such as accuracy, precision, recall, and F1-score. Evaluation results demonstrate that GraphSAGE consistently outperforms other models, achieving an accuracy of 95.80% on the DynKDD dataset, a notable improvement over the static NSL-KDD dataset. Other models follow a similar trend with a significant improvement on the constructed DynKDD dataset. The results validate the superiority of dynamic graph-based intrusion detection systems in identifying evolving network threats. With its ability to model evolving cyberattacks, DynKDD holds immense potential for real-world applications in protecting critical infrastructure from advanced and persistent threats, offering a transformative approach to network intrusion detection. © 2025 IEEE.

关键词： Network intrusion

来源：评论

学校读者我要写书评

暂无评论

Research on De-identification Applications of LLMs in Medical Records

引用

International Journal of Network Security 2025年第1期27卷 213-222页

作者： Hwang, Min-Shiang Fatima, Kanza Wang, Yang-Shu Wu, Nan-I Lin, Iuon-Chang Department of Computer Science & Information Engineering Asia University 500 Lioufeng Rd. Wufeng Taichung41354 Taiwan Department of Management Information Systems National Chung Hsing University Taiwan Department of Information Management Lunghwa University of Science and Technology Taoyuan33306 Taiwan

The primary objective of this research is to develop and promote efficient and secure de-identification technology to address the application of sensitive personal information in healthcare big data. Considering the outstanding performance of large language models (such as GPT) in understanding text, we aim to utilize these advanced models to process medical records. Compared to traditional methods such as manual processing, rule-based, or machine learning methods, large language models can more effectively de-identify unstructured medical record data. We can enhance the recognition and de-identification accuracy of personal identification information in medical records through deep learning techniques. Additionally, the contextual understanding capabilities of large language models can comprehensively process medical record data, ensuring that the de-identification process does not affect the integrity and availability of medical information. Therefore, the research methodology will include model training, optimization, and validation to ensure that the developed de-identification technology achieves efficient and reliable results in practical applications. We explore different large language models, considering the diversity of personal medical data to address various de-identification needs in different medical record data. In this article, we propose three research issues for de-identification: 1) Research on the application of deep language models and de-identification technology in medical cases;2) optimization of prompt engineering to enhance de-identification of medical cases by large language models;3) Optimization and promotion of de-identification technology for practical applications in the medical. © (2025), (Femto Technique Co., Ltd.). All rights reserved.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Exploring The Role of Interpretability in Adversarial Defense Mechanisms based Malware Classification

Exploring The Role of Interpretability in Adversarial Defens...

引用

Mobile Computing and Sustainable Informatics (ICMCSI), International Conference on

作者： t Assem Alhawari Sahar Ebadinezhad Department of Computer Information Systems Near East University Nicosia Cyprus Department of Computer Information Systems Computer Information Systems Research and Technology Centre (CISRTC) Near East University Nicosia Turkey

ISBN: (数字)9798331522667

ISBN: (纸本)9798331522674

This systematic literature review (SLR) presents a comprehensive overview of the current state of interpretability techniques in malware classification based on adversarial defense mechanisms, challenges, and future directions. Several ML and DL approaches have been integrated into the malware detection and classification defense systems, where adversarial attacks represent a crucial security threat against these defense systems. The interpretability techniques have been found very important to be applied to enhance the performance of these security systems and develop more explainable models and decisions for users and security experts. The SLR was conducted by combining and reviewing scientific literature on adversarial defense mechanisms-based interpretability techniques in the field of malware classification published between 2013 and 2024. This SLR presents a new insight into the categorization of interpretability techniques' implications in adversarial defense systems. We evaluate the importance and effectiveness of interpretability techniques by investigating ML models, adversarial attacks, evaluation methods, datasets, user trust, and malware features. Furthermore, the study provides challenges and future research directions for further existing issues

关键词： Analytical models Decision making Malware Informatics computer security Mobile computing Systematic literature review

来源：评论

学校读者我要写书评

暂无评论

Multi-Mode Inflow Prediction for Urban Transportation systems 14th

Multi-Mode Inflow Prediction for Urban Transportation System...

引用

14th International Conference on Frontier Computing, FC 2024

作者： Hung, Chih-Chieh Hung, Min-Hsien Chang, Jia-Wei Department of Management Information Systems National Chung Hsing University Taichung Taiwan Department of Computer Science and Information Engineering National Taichung University of Science and Technology Taichung Taiwan

ISBN: (纸本)9789819623945

Traffic congestion poses significant challenges to modern cities, leading to increased energy use, pollution, and long commute times. Optimizing public transit systems and encouraging their use is an effective solution. However, public transit integrates various modes like subways, buses, and shared bikes, requiring transfers between modes. Therefore, understanding the relationships between modes and accurately predicting passenger flows for each mode is crucial. Accurate predictions enable better capacity allocation, scheduling, and seamless transfers between modes, enhancing accessibility and convenience to attract more riders. Existing multi-modal prediction methods typically focus on station-to-region or region-to-region flows, overlooking station-to-station predictions that can better capture inter-modal relationships and provide fine-grained insights for traffic management. This study proposes a multi-modal framework for MRT and YouBike data that employs a spatial-temporal module to effectively model the spatiotemporal dependencies in the data. Evaluations on real-world datasets demonstrate the framework outperforms benchmark methods. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Traffic congestion

来源：评论

学校读者我要写书评

暂无评论

TMRE:Novel Algorithm for Computing Daily Reference Evapotranspiration Using Transformer-Based Models

引用

computers, Materials & Continua 2025年第5期83卷 2851-2864页

作者： Bushra Tayyaba Muhammad Usman Ghani Khan Talha Waheed Shaha Al-Otaibi Tanzila Saba National Center of Artificial Intelligence(NCAI) Al Khwarizmi Institute of Computer Sciences(KICS)Lahore54890Pakistan Department of Computer Science University of Engineering and Technology LahoreLahore54890Pakistan Artificial Intelligence&Data Analytics Lab Prince Sultan UniversityRiyadh11586Saudi Arabia Department of Information Systems College of Computer and Information SciencesPrincess Nourah bint Abdulrahman UniversityRiyadh11671Saudi Arabia

Reference Evapotranspiration(ETo)iswidely used to assess totalwater loss between land and atmosphere due to its importance in maintaining the atmospheric water balance,especially in agricultural and environmental *** estimation of ETo is challenging due to its dependency onmultiple climatic variables,including temperature,humidity,and solar radiation,making it a complexmultivariate time-series *** machine learning and deep learning models have been applied to forecast ETo,achieving moderate ***,the introduction of transformer-based architectures in time-series forecasting has opened new possibilities formore precise ETo *** this study,a novel algorithm for ETo forecasting is proposed,focusing on four transformer-based models:Vanilla Transformer,Informer,Autoformer,and FEDformer(Frequency Enhanced Decomposed Transformer),applied to an ETo dataset from the Andalusian *** novelty of the proposed algorithm lies in determining optimized window sizes based on seasonal trends and variations,which were then used with each model to enhance prediction *** custom window-sizing method allows the models to capture ETo’s unique seasonal patterns more ***,results demonstrate that the Informer model outperformed other transformer-based models,achievingmean square error(MSE)values of 0.1404 and 0.1445 for forecast windows(15,7)and(30,15),*** Vanilla Transformer also showed strong performance,closely following the *** findings suggest that the proposed optimized window-sizing approach,combined with transformer-based architectures,is highly effective for ETo *** novel strategy has the potential to be adapted in othermultivariate time-series forecasting tasks that require seasonality-sensitive approaches.

关键词： Reference evapotranspiration ETo transformer informer autoformer FEDformer timeseries forecasting self-attention

来源：评论

学校读者我要写书评

暂无评论

Using Local Gradient Compensation in Federated Learning for Improving Privacy and Performance 14th

Using Local Gradient Compensation in Federated Learning for ...

引用

14th International Conference on Frontier Computing, FC 2024

作者： Chang, Jia-Wei Chen, Wei-Cheng Hung, Chih-Chieh Department of Computer Science and Information Engineering National Taichung University of Science and Technology Taichung Taiwan Department of Management Information Systems National Chung Hsing University Taichung Taiwan

ISBN: (纸本)9789819623945

Federated learning (FL) has emerged as a leading approach for decentralized model training, preserving data privacy by exchanging only model parameters. However, recent studies have exposed vulnerabilities, revealing that attackers can reconstruct private data from shared model parameters, compromising user privacy. To address this, various defense methods have been proposed, with gradient perturbation being a lightweight yet effective strategy. However, it often incurs a trade-off with model performance. In this paper, we introduce a lightweight defense mechanism tailored for Federated Stochastic Gradient Descent (FedSGD). Our method leverages maximum gradient values to gauge noise scale, while also preserving and compensating for previous noise iterations to mitigate performance impacts. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

A Case Study of 3rd Party Hardware: The Weakest Link in Google’s Trustworthy Artificial Intelligence Implementation

引用

IEEE Access 2025年 13卷 93640-93648页

作者： Lorenz, Scott Stinehour, Stanley Chennamaneni, Anitha Subhani, Abdul B. Nadim, Mohammad Centex Technologies KilleenTX76541 United States Texas A&M University - Central Texas Department of Computer Information Systems TX76549 United States

The proliferated smart TV market has sparked a race among the tech giants to capture market share, with Google aggressively pursuing this domain through partnerships with third-party smart TV manufacturers. However, this expansion raises critical concerns regarding security vulnerabilities and the potential breakdown of Google’s Trustworthy Artificial Intelligence Implementation (TAII) standards. Despite the significant focus on software security, there is a lack of research on the security and privacy challenges associated with smart TV hardware. In this paper, we present a case study exploring the security vulnerabilities present in TCL Smart TVs integrated with Google Cloud services. Our case study result identifies significant risks of consumer data breaches and exposure of personal information. Our findings show significant gaps in following the TAII principles of privacy, security, safety, and transparency. © 2013 IEEE.

关键词： Sensitive data

来源：评论

学校读者我要写书评

暂无评论

On the Exponential Growth of Geometric Shapes 20th

On the Exponential Growth of Geometric Shapes

引用

20th International Symposium on Algorithmics of Wireless Networks, ALGOWIN 2024

作者： Almalki, Nada Gupta, Siddharth Michail, Othon Department of Computer Science University of Liverpool Liverpool United Kingdom Department of Computer Science and Information Systems BITS Pilani K K Birla Goa Campus Goa India

ISBN: (纸本)9783031745799

In this paper, we explore the exponential growth of geometric structures starting from a single node, focusing on centralized growth operations. We identify a parameter k, representing the number of turning points within specific parts of a shape. We prove that, if edges can only be formed when generating new nodes and cannot be deleted, trees having at most k turning points on every root-to-leaf path can be grown in O(klogn) time steps and spirals with O(logn) turning points can be grown in O(logn) time steps, n being the size of the final shape. For this model, we also show that the maximum number of turning points in a root-to-leaf path of a tree is a lower bound on the number of time steps to grow the tree and that there exists a class of paths such that any path in the class with k turning points requires Ω(klogk) time steps to be grown. If nodes can additionally be connected as soon as they become adjacent, we prove that if a shape S has a spanning tree with at most k turning points on every root-to-leaf path, then the adjacency closure of S can be grown in O(klogn) time steps. In the strongest version of the model, where, additionally, edges can be deleted and neighbors handed over to new nodes, we present a universal algorithm for growing any shape S exponentially fast. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Trees (mathematics)

来源：评论

学校读者我要写书评

暂无评论

Augmented Karuta: Interactive Playing Cards on the Floor for Learning Local Culture 7

Augmented Karuta: Interactive Playing Cards on the Floor for...

引用

7th IEEE International Conference on Artificial Intelligence and eXtended and Virtual Reality, AIxVR 2025

作者： Umezu, Nobuyuki Bando, Hideki Ibaraki University Dept. of Computer and Information Sciences Hitachi Japan Ibaraki University Dept. of Mechanical Systems Engineering Hitachi Japan

ISBN: (纸本)9798331521578

This paper introduces Augmented Karuta, an interactive floor projection experience inspired by traditional Japanese playing cards, known as 'Karuta'. Designed to engage users and stimulate interest in local cultures, history, and traditions, the system utilizes three projectors. Two projectors project 46 'grabbing' cards onto the floor screen of 4.1 x 3.7 (m), while the third displays a corresponding 'reading' card on a wall screen. Each matching pair of cards shares a poem or sentence that begins with one of the Japanese phonograms. During a session, users actively engage by stepping onto the projected 'grabbing' card that matches the 'reading' card displayed on the wall, effectively blending physical movement with cognitive skills to find the correct match. By integrating a two-dimensional LiDAR sensor, the system tracks the user's position on the interactive floor. This tracking is vital for the proposed system to immediately respond to the user's actions, ensuring an immersive and seamless interaction with the digital karutas on the floor. We conducted evaluation experiments at two museums, gathering positive feedback from 57 participants. Future work will focus on quantitatively evaluating the system's impact on learning effectiveness. © 2025 IEEE.

关键词： Tropics

来源：评论

学校读者我要写书评

暂无评论

Assessing Personalized AI Mentoring with Large Language Models in the Computing Field

Assessing Personalized AI Mentoring with Large Language Mode...

引用

2025 IEEE Symposium on Computational Intelligence in Natural Language Processing and Social Media, CI-NLPSoMe Companion 2025

作者： Luo, Xiao O'Connell, Sean Mithun, Shamima Oklahoma State University Department of Management Science and Information Systems United States Purdue University Department of Computer Information Technology Indianapolis United States

ISBN: (纸本)9798331519742

This paper provides an in-depth evaluation of three state-of-the-art Large Language Models (LLMs) for personalized career mentoring in the computing field, using three distinct student profiles that consider gender, race, and professional levels. We evaluated the performance of GPT-4, LLaMA 3, and Palm 2 using a zero-shot learning approach without human intervention. A quantitative evaluation was conducted through a custom natural language processing analytics pipeline to highlight the uniqueness of the responses and to identify words reflecting each student's profile, including race, gender, or professional level. The analysis of frequently used words in the responses indicates that GPT-4 offers more personalized mentoring compared to the other two LLMs. Additionally, a qualitative evaluation was performed to see if human experts reached similar conclusions. The analysis of survey responses shows that GPT-4 outperformed the other two LLMs in delivering more accurate and useful mentoring while addressing specific challenges with encouragement languages. Our work establishes a foundation for developing personalized mentoring tools based on LLMs, incorporating human mentors in the process to deliver a more impactful and tailored mentoring experience. © 2025 IEEE.

关键词： Natural language processing systems

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：