检索结果-内蒙古大学图书馆

14th International Conference on Ubiquitous and Future Networks, ICUFN 2023

作者： Shan, Gaoyang Choi, Geunkyung Roh, Byeong-Hee Ajou University Dept. of Software and Computer Engineering Suwon16499 Korea Republic of Ajou University Dept. of Ai Convergence Nework Dept. of Computer Engineering Suwon16499 Korea Republic of

ISBN: (纸本)9798350335385

To adapt to various IoT services, Bluetooth Low Energy (BLE) specification has defined a variety of communication modes, such as advertising, extended advertising, and periodic advertising events. As services' environments, such as the number of users and QoS requirements, are changed, modes of protocols should be also adjusted to meet those services' needs. This paper proposes an adaptive BLE scheme to reflect dynamically changing services' environments. The effectiveness of the proposed scheme is verified through simulation experiments. © 2023 IEEE.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

GolfPoseNet: Golf-Specific 3D Human Pose Estimation Network

GolfPoseNet: Golf-Specific 3D Human Pose Estimation Network

引用

2025 International Conference on Electronics, Information, and Communication, ICEIC 2025

作者： Park, BoJeong Kim, JongWoo Mun, SeoYeong Choi, Younglim Kim, Hyunseok Dong-A University Dept. of Electric Engineering Busan Korea Republic of Dong-A University Dept. of Software Innovation Center Busan Korea Republic of Dong-A University Dept. of Computer Engineering Busan Korea Republic of

ISBN: (纸本)9798331510756

This paper aims to develop a robotic golf trainer using a wheeled robot and create a model that accurately recognizes the user's golf motion. Since existing 3D pose models show limitations in golf motion recognition [1], [2], we propose an optimized recognition model by combining various methods and preprocessing golf datasets to overcome these problems. We used HRNet to estimate 2D pose from video frames and fed these data into a transformer-based model to generate three hypotheses through residual linking and embedding layers. These were integrated to finally design a structure to predict 3D pose, and batch normalization, dropout, and LeakyReLU were used to improve the stability and nonlinearity handling of the model. Experimental results show that the proposed model effectively learns temporal information from multi-view 2D pose data without 3D annotation and performs well even when 3D data is scarce. The proposed method can be used in various applications such as human-robot interaction and sports motion analysis including golf, and further research on multiple datasets and real-world environments is expected to improve further and generalize the model's performance. © 2025 IEEE.

关键词： Human robot interaction

来源：评论

学校读者我要写书评

暂无评论

Macula Swelling Detection using Retinal Fundus Images

Macula Swelling Detection using Retinal Fundus Images

引用

2023 IEEE International Conference on Imaging Systems and Techniques, IST 2023

作者： Khan, Rukham Mumtaz Ali Khan, Fazal Saadat Khan, Fauzia Hassan, Rafia National University of Sciences and Technology Dept. of Computer Engineering Islamabad Pakistan University of Tartu Dept. of Computer Science Tartu Estonia Bahria University Dept. of Software Engineering Islamabad Pakistan

ISBN: (纸本)9798350330830

Diabetic Macular Edema (DME) is an advanced symptom of Diabetic Retinopathy (DR) and can lead to irreversible vision loss. Macular Edema (ME) is the swelling of the macula, the part of the eye accountable for thorough central vision. The utilization of image processing in the detection of diabetic macular edema (DME) plays a crucial role in the field of medical imaging, effectively reducing the time and manual effort required by medical specialists to identify this disease. In this paper, we proposed a method that uses digital fundus images to analyze the swelling of the macula. The proposed algorithm consists of four steps, i.e., (i) retinal image dataset collection, (ii) retinal image pre-processing, (iii) feature extraction, and (iv) classification to categorize the disease into three different stages, i.e., normal, stage 1 (mild) and stage 2 (severe). The proposed method is tested on three databases, i.e., MESSIDOR, DMED, and the local dataset AFIO obtained from the local hospital in Pakistan. We use three parameters to check the validity of the proposed algorithm, i.e., visual inspection, accuracy, and area under calculation time. Our method achieves an average accuracy of 96.07 % with a sensitivity of 93.74 % and specificity of 97.32%. We also compare our results with other proposed methods, which shows our method performs better. © 2023 IEEE.

关键词： Eye protection

来源：评论

学校读者我要写书评

暂无评论

Multi-Label Classification of CS Papers Using Natural Language Processing Models 22

Multi-Label Classification of CS Papers Using Natural Langua...

引用

22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

作者： Zhang, Zheng Das, Amit Rahgouy, Mostafa Bao, Yida Baskiyar, Sanjeev Auburn University Dept. of Computer Science & Software Engineering Dept. of Mathematics & Statistics AuburnAL United States

ISBN: (纸本)9798350345346

computer science (CS) is a rapidly evolving field of knowledge with a significant volume and significance of scientific papers. Consequently, meticulous management and categorization of CS papers are crucial to ensure efficient access and retrieval of relevant information. This study aims to evaluate the effectiveness of eight traditional machine learning models, as well as Recurrent Neural Network (RNN) models, such as LSTM and BiLSTM, and state-of-the-art pre-trained self-attention-based natural language processing (NLP) models, including BERT, XLNet, RoBERTa, and DistilBERT, for the task of multi-label classification of arXiv CS paper classification. The results demonstrate that pre-trained self-attention-based models consistently outperform the other models regarding classification performance. Moreover, self-attention-based models exhibited superior performance and achieved a new state-of-the-art result. To the best of our knowledge, our paper represents the first endeavor to evaluate machine learning models, specifically on computer science-related documents, as part of a multi-label classification task. © 2023 IEEE.

关键词： BERT Deep Learning LSTM Machine learning multi-label classification

来源：评论

学校读者我要写书评

暂无评论

Towards the Diagnosis of Heart Disease Using an Ensemble Learning Approach 22

Towards the Diagnosis of Heart Disease Using an Ensemble Lea...

引用

22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

作者： Zhang, Zheng Xu, Liangliang Bao, Yida Baskiyar, Sanjeev Auburn University Dept. of Computer Science & Software Engineering Dept. of Mathematics & Statistics AuburnAL United States

ISBN: (纸本)9798350345346

Heart disease is one of the leading causes of death worldwide. Understanding the presence of heart disease is crucial for timely intervention and effective management. However, it is still a challenging task to accurately diagnose and treat heart disease. In this paper, we propose an ensemble machine learning model that combines the predictive power of three state-of-the-art machine learning algorithms: BERT (Bidirectional Encoder Representations from Transformers), FT-transformer, and XG-Boost. Through extensive training and evaluation, we assess the performance of our model using established metrics such as accuracy, precision, recall, F1-score, and receiver operating char-acteristic (ROC) curve analysis. The results obtained demonstrate the effectiveness and validity of our machine learning system in predicting heart disease accurately. This paper presents a promising approach that can assist healthcare professionals in making informed decisions and improving patient outcomes, enhances our understanding of heart disease patterns, and contributes to the development of effective diagnostic and treatment approaches. © 2023 IEEE.

关键词： Autoencoder BERT Heart Disease Machine Learning Transformer XG-Boost

来源：评论

学校读者我要写书评

暂无评论

MTDC Fault Detection and Localization Using High-Pass Filter 50

MTDC Fault Detection and Localization Using High-Pass Filter

引用

50th Annual Conference of the IEEE Industrial Electronics Society, IECON 2024

作者： Khan, Saad Ahmed Ukil, Abhisek Nair, Nirmal-Kumar C. Li, Dongyu Dept. of Electrical Computer and Software Engineering The University of Auckland Auckland New Zealand School of Electrical and Automation Engineering East China Jiaotong University Nanchang330013 China

ISBN: (纸本)9781665464543

The DC link capacitor discharge in the event of a DC fault is rapid and contains high frequencies. This rapid discharge of high current and frequency interface with DC bus, Voltage Source Converter (VSC), and AC source. This interface damages the equipment and possibly living beings in proximity. Therefore, it is necessary to develop a topology to detect and identify fault types promptly, along with the isolation and restoration of the system. This study introduces an improved novel fault detection technique using the Highpass Chebyshev type 2 filter due to its flatter pass-band response. Further, the polarities of peaks obtained from the fault detection calculation are used again for another proposed novel fault location method. It identifies fault types to locate and isolate the faulty system using only polarities of amplitude response peaks. In the simulation, both methods were fast and accurate in detecting, locating, and identifying appropriate fault types only by High Pass Filter (HPF) amplitude response peaks and their polarities. © 2024 IEEE.

关键词： Chebyshev filters

来源：评论

学校读者我要写书评

暂无评论

Exploring the Potential of Llama Models in Automated Code Refinement: A Replication Study

arXiv

引用

arXiv 2024年

作者： Caumartin, Genevieve Qin, Qiaolin Chatragadda, Sharon Panjrolia, Janmitsinh Li, Heng Costa, Diego Elias Dept of Computer Science and Software Engineering Concordia University Montreal Canada Dept of Computer Engineering and Software Engineering Polytechnique Montreal Montreal Canada

Code reviews are an integral part of software development and have been recognized as a crucial practice for minimizing bugs and favouring higher code quality. They serve as an important checkpoint before committing code and play an essential role in knowledge transfer between developers. However, code reviews can be time-consuming and can stale the development of large software projects. In a recent study, Guo et al. assessed how ChatGPT3.5 can help the code review process. They evaluated the effectiveness of ChatGPT in automating the code refinement tasks, where developers recommend small changes in the submitted code. While Guo et al.’s study showed promising results, proprietary models like ChatGPT pose risks to data privacy and incur extra costs for software projects. In this study, we explore alternatives to ChatGPT in code refinement tasks by including two open-source, smaller-scale large language models: CodeLlama and Llama 2 (7B parameters). Our results show that, if properly tuned, the Llama models, particularly CodeLlama, can achieve reasonable performance, often comparable to ChatGPT in automated code refinement. However, not all code refinement tasks are equally successful: tasks that require changing existing code (e.g., refactoring) are more manageable for models to automate than tasks that demand new code. Our study highlights the potential of open-source models for code refinement, offering cost-effective, privacy-conscious solutions for real-world software development. Copyright © 2024, The Authors. All rights reserved.

关键词： Reviews

来源：评论

学校读者我要写书评

暂无评论

Unveiling the Arsenal of User Data Protection Tools and Practices 1

Unveiling the Arsenal of User Data Protection Tools and Prac...

引用

1st IEEE Karachi Section Humanitarian Technology Conference, Khi-HTC 2024

作者： Moiz, Syed Abro, Abdul Ahad Ebrahim, Mansoor Abro, Abdul Basit Iqra University Main Campus Dept. of Computer Science Karachi Pakistan Iqra University Main Campus Dept. of Software Engineering Karachi Pakistan Indus University Dept. of Computer Science Karachi Pakistan

ISBN: (纸本)9798350373042

In our rapidly evolving digital landscape, the imperative of safeguarding personal data has surged in significance. As lives increasingly intertwine with digital technologies, personal information has grown markedly, parallel to the escalating threats posed by cyberattacks and data breaches. This research paper provides a comprehensive exploration of data security and privacy, aiming to impart a profound understanding of the intrinsic value of data. It meticulously examines essential methodologies such as encryption, access control, multifactor authentication, and systems, all of which collectively form the bedrock of a resilient security framework. Additionally, this study delves into the intricate regulatory landscape that governs data protection, placing particular emphasis on well-established frameworks such as the General Data Protection Regulation and the Health Insurance Portability and Accountability Act. It highlights emergent trends in data security, including the integration of artificial intelligence and machine learning for threat detection, while also addressing the distinctive challenges posed by the proliferation of Internet of Things (IoT) devices. In conclusion, this paper underscores the paramount importance of personal data protection in our digital age, offering invaluable insights and guidance for organizations and individuals endeavoring to bolster their defenses and uphold data integrity within our interconnected world. © 2024 IEEE.

关键词： Data privacy

来源：评论

学校读者我要写书评

暂无评论

Fingerprint generation and authentication though Adaptive convolution generative adversarial network (ADCGAN) 7

Fingerprint generation and authentication though Adaptive co...

引用

7th International Multi-Topic ICT Conference: IMTIC 2023

作者： Mustafa, Syed Muhammad Nabeel Zehra, Syeda Sundus Baber, Alina Siddiqui, Maria Andleeb Ned University of Engineering and Technology Dept. of Computer Science & Information Technology Karachi Pakistan Sir Syed University of Engineering and Technology Dept of Software Engineering Karachi Pakistan

ISBN: (纸本)9798350338461

Fingerprints are crucial in identification of humans. The uniqueness of finger prints makes it an interesting subject. Fingerprints are termed as a technique used to define, assess, and quantify a person's physical and behavioral property. Deep learning has made its application in all the major fields such as natural language processing, computer vision and speech processing. Deep learning has also found its application in the important subject of fingerprint synthesis and biometric. The ever-growing complexity of fingerprint authentication issues, from cellphone authentication to airport security systems, seems to be best handled by these models. In recent years, deep learning-based models have been used more and more to raise the accuracy of various fingerprint recognition systems. The persuasive capacity of Generative Adversarial Networks (GANs) to generate believable instances can be credibly taken from an existing distribution of samples. GAN exhibits exceptional performance on data generation-based tasks and also encourages study in privacy and security. In this work, using Adaptive Deep Convolution Generative Adversarial Networks (ADCGAN), we develop a model that generates and authenticate the fingerprints. A Socofing dataset was trained on ADGAN model. The model gave 92% accuracy. The conduct of fingerprint research has been made possible due to ADGAN, without restrictions related to the confidential nature of biometric data. © 2023 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Using Deep Learning depth Maps to Improve Monocular SLAM 38

Using Deep Learning Depth Maps to Improve Monocular SLAM

引用

38th International Conference on Image and Vision Computing New Zealand, IVCNZ 2023

作者： Hollows, Simon Green, Richard University of Canterbury Dept. Mechatronics Engineering Christchurch New Zealand University of Canterbury Dept. Computer Science and Software Engineering Christchurch New Zealand

ISBN: (纸本)9798350370515

This paper proposes the use of a neural network to generate depth maps from 2D frames, offering an alternative to depth cameras. The goal is to address the accuracy limitations of monocular SLAM by gaining more accurate depth data through deep learning. This approach combines the simplicity and cost-effectiveness of a single camera setup with the enhanced accuracy of RGB-D ***-SLAM running with RGB-D input was used as the SLAM framework for this method. This allowed for accurate comparisons between depth sensor depth maps in test datasets and the generated depth maps. The MiDaS neural network was used to generate estimated depth maps. These were then scaled to create consistency over the entire test sequence. The TUM RGB-D dataset was used for *** generated depth maps were able to compare with the TUM RGB-D data in accuracy. On one dataset, the TUM depth maps had a root mean square error (RMSE) or 0.24 m while the generated depth maps had a RMSE of 0.18 m. In general, the proposed method lacks consistency and the lack of good fusion between depth maps causes this approach to fail in slow-moving scenes. © 2023 IEEE.

关键词： Cost effectiveness

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：