检索结果-内蒙古大学图书馆

Federated Feature Concatenate Method for Heterogeneous Computing in Federated Learning

computers, Materials & Continua 2023年第4期75卷 351-371页

作者： Wu-Chun Chung Yung-Chin Chang Ching-Hsien Hsu Chih-Hung Chang Che-Lun Hung Department of Information and Computer Engineering Chung Yuan Christian UniversityTaoyuanTaiwan Department of Computer Science and Information Engineering Asia UniversityTaichungTaiwan Department of Medical Research China Medical University HospitalChina Medical UniversityTaichungTaiwan Department of Computer Science&Communication Engineering Providence UniversityTaichungTaiwan Institute of Biomedical Informatics National Yang Ming Chiao Tung UniversityTaipeiTaiwan

Federated learning is an emerging machine learning techniquethat enables clients to collaboratively train a deep learning model withoutuploading raw data to the aggregation server. Each client may be equippedwith different computing resources for model training. The client equippedwith a lower computing capability requires more time for model training,resulting in a prolonged training time in federated learning. Moreover, it mayfail to train the entire model because of the out-of-memory issue. This studyaims to tackle these problems and propose the federated feature concatenate(FedFC) method for federated learning considering heterogeneous *** leverages the model splitting and feature concatenate for offloadinga portion of the training loads from clients to the aggregation server. Eachclient in FedFC can collaboratively train a model with different cutting ***, the specific features learned in the deeper layer of the serversidemodel are more identical for the data class classification. Accordingly,FedFC can reduce the computation loading for the resource-constrainedclient and accelerate the convergence time. The performance effectiveness isverified by considering different dataset scenarios, such as data and classimbalance for the participant clients in the experiments. The performanceimpacts of different cutting layers are evaluated during the model *** experimental results show that the co-adapted features have a criticalimpact on the adequate classification of the deep learning model. Overall,FedFC not only shortens the convergence time, but also improves the bestaccuracy by up to 5.9% and 14.5% when compared to conventional federatedlearning and splitfed, respectively. In conclusion, the proposed approach isfeasible and effective for heterogeneous clients in federated learning.

关键词： Federated learning deep learning artificial intelligence heterogeneous computing

来源：评论

学校读者我要写书评

暂无评论

Tailoring Fano resonance characteristics for GSS4T1-based reconfigurable photonics

Tailoring Fano resonance characteristics for GSS4T1-based re...

引用

2024 Frontiers in Optics, FiO 2024

作者： Foteinopoulou, Stavroula Tsitsas, Nikolaos L. Department of Electrical and Computer Engineering University of New Mexico NM87131-0001 United States School of Informatics Aristotle University of Thessaloniki Thessaloniki54124 Greece

We present a GSS4T1-based metagrating designed to exhibit a homogeneous-optical-medium response, when GSS4T1 amorphous, or a judicious Fano response, when GSS4T1 crystalline, which enables a negative beam steering tha... 详细信息

关键词： Photonics

来源：评论

学校读者我要写书评

暂无评论

Effect of CLAHE-based Enhancement on Bean Leaf Disease Classification through Explainable AI 12

Effect of CLAHE-based Enhancement on Bean Leaf Disease Class...

引用

12th IEEE Global Conference on Consumer Electronics, GCCE 2023

作者： Noviandy, Teuku Rizky Maulana, Aga Khowarizmi, Faris Muchtar, Kahlil Universitas Syiah Kuala Artificial Intelligence Department of Informatics Aceh Indonesia Universitas Syiah Kuala Department of Electrical and Computer Engineering Aceh Indonesia

ISBN: (纸本)9798350340181

Global agriculture faces a major threat from plant diseases, particularly those affecting bean leaves, resulting in significant crop yield losses and impacting farmers' livelihoods and food production. To overcome the limitations of manual visual inspection methods, researchers have adopted computer vision and deep learning techniques for automated disease detection. This study improves the accuracy of classifying bean leaf diseases using Contrast Limited Adaptive Histogram Equalization (CLAHE) and the EfficientNetB0 architecture. Explainable AI technique, specifically Grad-CAM, is employed to analyze predicted images for transparency and interpretability. The proposed approach outperforms models trained on original images, achieving higher accuracy, precision, recall, and F1 score. CLAHE proves to be effective in optimizing CNN-based models for plant disease detection. Grad-CAM analysis reveals the model's focus on disease-related features, enhancing understanding of its behavior. This research contributes to the development of robust automated systems for plant disease detection, benefiting farmers and agricultural practitioners in decision-making. © 2023 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Causal Effects of Environmental Exposures and Biological Traits on the Difference between Phenotypic and Chronological Ages

Causal Effects of Environmental Exposures and Biological Tra...

引用

2023 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2023

作者： Pala, Daniele Xie, Yuezhi Xu, Jia Zhang, Yuqin Shen, Li University of Pavia Department of Computer Electrical and Biomedical Engineering Pavia Italy University of Pennsylvania Department of Biostatistics Epidemiology and Informatics Philadelphia United States

ISBN: (纸本)9798350337488

Aging is a physiological process associated with numerous cardiovascular, degenerative and neurological conditions. The current demographic changes in the world, characterized by an increasing average age especially in Europe and North America, are causing an increment in the prevalence of such conditions, leading to a public health problem as more resources are required to manage treatments. Recent research has defined a new score named Phenotypic Age, i.e. an adjusted age that takes into account the current health status considering a series of biomarkers, that can be useful to provide a more precise estimation of the probability of developing aging-related conditions. Prevention of such conditions can be performed more efficiently by studying the mechanisms that lead to an increased phenotypic age rather than attempting to treat a patient that has already a high health risk. In this work, we combine pairwise association techniques and mediation analysis to define a strategy to investigate the inner causal mechanisms that lead from specific environmental exposures to an increasing gap between phenotypic and chronological age, considering the influence of biological variables. Four environmental exposures and 11 biological traits have been identified in the NHANES dataset, and each trait has been tested as a mediation variable for each exposure. Almost all mediations reported significant indirect effects with specific results that provide new insights into the causal mechanisms that lead to a deranged aging rate. © 2023 IEEE.

关键词： Aging Exposome Mediation Analysis Phenotypic Age Public Health

来源：评论

学校读者我要写书评

暂无评论

Wireless Edge Controller for Robot Education Using MQTT on Dobot Magician Lite 1

Wireless Edge Controller for Robot Education Using MQTT on D...

引用

1st IEEE International Symposium on Consumer Technology, ISCT 2024

作者： Prayudi, Agus Arief, Muhammad Fauzan Zainuddin, Muhammad Agus Sukaridhoto, Sritrusta Rante, Hestiasari Politeknik Elektronika Negeri Surabaya Department of Electrical Engineering Surabaya Indonesia Politeknik Elektronika Negeri Surabaya Department of Creative Multimedia Technology Surabaya Indonesia Politeknik Elektronika Negeri Surabaya Department of Informatics and Computer Engineering Surabaya Indonesia

ISBN: (纸本)9798350365191

The integration of wireless controllers in robot education has emerged as a pivotal area to enhance the learning experience and practical understanding of robotics concepts. This paper presents the implementation of a wireless controller for educational purposes using MQTT (Message Queuing Telemetry Transport) on the Dobot Magician Lite robotic platform. MQTT, a lightweight messaging protocol, facilitates efficient communication between the controller and the robot, enabling seamless interaction and control over the robot's movements and functionalities. The implementation focuses on providing a user-friendly interface and intuitive controls, allowing students and educators to easily operate the robot while exploring various educational scenarios. Leveraging MQTT's flexibility, the system offers real-time control and monitoring capabilities, fostering interactive learning experiences in robotics education. The proposed wireless controller demonstrates the feasibility and effectiveness of employing MQTT in robot education, paving the way for enhanced learning opportunities and practical applications in the field of robotics. The examination of MQTT performance across various bandwidths provides important insights into our system's abilities. At 8 Mbps, our system achieved its highest transaction speed, averaging 1.61 transactions per second with a quick average transaction time of 0.62 seconds. Transaction speeds decreased as bandwidth decreased, with 1 Mbps showing the slowest average TPS of 1.24. Importantly, data retention stayed at 100% for all bandwidths, demonstrating consistent reliability. © 2024 IEEE.

关键词： Educational robots

来源：评论

学校读者我要写书评

暂无评论

A Tutorial on Hybrid Feedback Control 63

A Tutorial on Hybrid Feedback Control

引用

63rd IEEE Conference on Decision and Control, CDC 2024

作者： Casau, Pedro Ferrante, Francesco Sanfelice, Ricardo G. University of Aveiro Department of Electronics Telecommunications and Informatics Portugal University of Perugia Department of Engineering Italy University of California Faculty of Electrical and Computer Engineering Santa Cruz United States Instituto de Telecomunicações Portugal

ISBN: (纸本)9798350316339

This tutorial paper introduces hybrid feedback control through a self-contained examination of hybrid control systems modeled by the combination of differential and difference equations with constraints. Using multiple examples, it illustrates the power of hybrid feedback control, which stems from the integration of continuous and discrete dynamics, where state variables update instantaneously at specific events while flowing continuously otherwise. The paper defines hybrid closed-loop systems as interconnected hybrid plants and controllers with designated inputs and outputs, and formalizes their solutions. It summarizes key properties of hybrid systems and reviews various control strategies, including supervisory control with logic variables to select feedback controllers, event-triggered control to minimize control input updates, and strategies using multiple Lyapunov-like functions for stabilization. Pointers to further reading and other strategies in the literature are provided. © 2024 IEEE.

关键词： Feedback control

来源：评论

学校读者我要写书评

暂无评论

Assessing the Effectiveness of Oversampling and Undersampling Techniques for Intrusion Detection on an Imbalanced Dataset 4

Assessing the Effectiveness of Oversampling and Undersamplin...

引用

4th IEEE Industrial Electronics and Applications Conference, IEACon 2023

作者： Rahma, Fayruz Rachmadi, Reza Fuad Pratomo, Baskoro Adi Purnomo, Mauridhi Hery Institut Teknologi Sepuluh Nopember Department of Electrical Engineering Surabaya Indonesia Universitas Islam Indonesia Department of Informatics Yogyakarta Indonesia Institut Teknologi Sepuluh Nopember Department of Computer Engineering Surabaya Indonesia Institut Teknologi Sepuluh Nopember Department of Informatics Surabaya Indonesia

ISBN: (纸本)9798350347517

The imbalanced class distribution in intrusion detection systems has been a significant issue. Imbalanced class distribution can negatively impact the performance of intrusion detection systems as they may be biased towards the majority class. We explore the effectiveness of oversampling and under-sampling techniques to address this issue. Oversampling and undersampling techniques aim to balance the class distribution and improve the performance of the intrusion detection system. Oversampling increases the number of records in the minority class to make it closer in size to the majority class. Conversely, undersampling reduces the number of records in the majority class so that it is closer in size to the minority class. We assess the effectiveness of different oversampling and undersampling techniques, including Random OverSampling, SMOTE, ADASYN, Random UnderSampling, AllKNN, TomekLinks, SMOTEENN, and SMOTETomek. The experiment's findings indicate that the raw data achieved the highest accuracy score, 0.965. On the other hand, the Random Oversampling method yielded the highest F1 score, reaching a score of 0.589. When we see the evaluation scores of each class, the recall & F1 scores generally show high contrast between classes with a large amount of data and classes with (previously) a small amount of data, even though the data for training has been more balanced. We found that oversampling and undersampling can improve the performance of intrusion detection systems in specific ways, but this still needs improvement. These results can serve as a reference for researchers developing intrusion detection systems. © 2023 IEEE.

关键词： Intrusion detection

来源：评论

学校读者我要写书评

暂无评论

Development of IoT-Healthcare Model for Gastric Cancer from Pathological Images 1

引用

9th International Conference on Mining Intelligence and Knowledge Exploration, MIKE 2023

作者： Belgaum, Mohammad Riyaz Momina, Shaik Maryam Farhath, L. Nousheen Nikhitha, K. Naga Jyothi, K. Department of Computer Science and Engineering G. Pullaiah College of Engineering and Technology Kurnool India Faculty of Computing and Informatics Multimedia University Cyberjaya Malaysia

ISBN: (数字)9783031440847

ISBN: (纸本)9783031440830

The diagnosis of stomach cancer automatically in digital pathology images is a difficult problem. Gastric cancer (GC) detection and pathological study can be greatly aided by precise region-by-region segmentation. On a technical level, this issue is complicated by the fact that malignant zones might be any size or shape and have fuzzy boundaries. The research employs a deep learning-based approach and integrates many bespoke modules to cope with these issues. The channel refinement model is the attentional actor on the chin channel. While implementing the feature channel, the learnt channel weight can be used to eliminate unnecessary features. Calibration is essential to improve classification precision. The results of channel recalibration can be improved with the help of a re-calibration (MSCR) model. The top pooling layer of the network is where the multiscale attributes are sent. The outcomes of channel recalibration may be enhanced by using the channel weights found at various scales as input to the next channel recalibration perfect. Our unique gastric cancer segmentation dataset, carefully glossed down to the pixel level by medical authorities, is used for extensive experimental comparisons. The numerical comparisons with other approaches show that our strategy is superior. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Diagnosis

来源：评论

学校读者我要写书评

暂无评论

Explainable Emotion Recognition in Social Networks with Transformers

Explainable Emotion Recognition in Social Networks with Tran...

引用

International Conference on computer and Information Science (ACIS)

作者： Isidoros Perikos Chara Pylarinou Computer Engineering and Informatics Department University of Patras Computer Technology Institute and Press “Diophantus” Patras Greece Computer Engineering and Informatics Department University of Patras Patras Greece

ISBN: (数字)9798350376647

ISBN: (纸本)9798350376654

Emotion recognition in text has become an essential research area within artificial intelligence and natural language processing due to its applications in sentiment analysis, human-computer interaction, and social media analysis. This paper introduces an approach for emotion recognition in social networks using transformer models enhanced with explainability techniques. By leveraging the advanced capabilities of transformers to analyze and classify emotions in written content, our finetuned models achieves high accuracy in identifying emotions according to the Ekman emotional model. The integration of LIME and SHAP provides transparency and interpretability to our model, making it more trustworthy for practical applications. Our findings indicate that models like XLNET report an exceptional performance in emotion recognition tasks. The explainability techniques illustrate the transformer model decision-making processes and the way that they utilize words and their continual information in sentences.

关键词： Human computer interaction Emotion recognition Analytical models Sentiment analysis Social networking (online) Computational modeling Predictive models Transformers Data models Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

Implementation of the SPECK Cipher Algorithm on Patient Blood Pressure Data Using the Internet of Things (IoT) 7

Implementation of the SPECK Cipher Algorithm on Patient Bloo...

引用

7th International Conference on Sustainable Information engineering and Technology, SIET 2022

作者： Santoso, Edy Nurwarsito, Heru Malik, Argandhi Adi Department of Informatics Engineering Faculty of Computer Science University of Brawijaya Malang Indonesia

ISBN: (纸本)9781450397117

Internet of Things technology allows humans to collect information about daily activities, for example to find out someone's health information, but at the time of data transmission it has an impact on one's privacy. Sniffing can record data sent from the user to the database server, this results in a leak of individual privacy. The SPECK algorithm is one of the cryptographic algorithms that guarantees confidentiality because the SPECK algorithm has varying key lengths and block sizes. Adding the SPECK cipher Algorithm to the node that holds sensor data can minimize privacy data leaks caused by data transmission on the Internet of Things. The SPECK algorithm has the same results as the test vectors in published journals and the resulting ciphertext can be returned to its original form by decryption. The encryption time is 0.089354 seconds and the decryption time on the server is 0.00610295 seconds. © 2022 ACM.

关键词： Internet of things

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：