Federated optimization or federated learning (FL) involves optimization of the global model or the server model by minimizing the global loss function which is weighted average of all the local loss functions. The opt...
Federated optimization or federated learning (FL) involves optimization of the global model or the server model by minimizing the global loss function which is weighted average of all the local loss functions. The optimization of the global model requires faster convergence to reduce the number of communication rounds or global iterations which is one of the major challenge in federated optimization. This paper propose FONN which handles this communication overhead in federated optimization by utilizing Nys-Newton, while updating local models. As compared to existing state-of-the-art FL algorithms, SCAFFOLD, GIANT and DONE, utilization of Nys-Newton leads to better convergence and reduction in communication rounds or global iterations while achieving a desired performance from the global model which may be observed from the experimental results on various heterogeneously partitioned datasets.
The Internet plays a significant part in the digital world. With the expanding use of web and online applications, different ventures go to huge information answers for adapt up to the evolving request. In this large ...
ISBN:
(数字)9798350302820
ISBN:
(纸本)9798350302837
The Internet plays a significant part in the digital world. With the expanding use of web and online applications, different ventures go to huge information answers for adapt up to the evolving request. In this large information period, information is the most important element as it is mainly utilized by organizations for future expectation and decision making. The internet has a seriously advanced a clever way for individuals to communicate their perspectives and suppositions about various themes, patterns and issues. The opinion examination recognized the sensations of client about a few items, subjects into three preset classes: positive, negative or impartial alongside the issue of mining. This paper presents a study on big information analytics on twitter opinion in cloud climate and how huge information controls a lot of information. This paper presents a concise investigation of methodologies which have been utilized in huge information analytics on twitter opinion in cloud climate.
DC microgrids, which utilize PV arrays and batteries or wind turbines and batteries, offer a viable solution for meeting the low to medium-power needs of rural, remote, and informal settlements. However, these solutio...
DC microgrids, which utilize PV arrays and batteries or wind turbines and batteries, offer a viable solution for meeting the low to medium-power needs of rural, remote, and informal settlements. However, these solutions have limitations since they rely on the availability of the sun or wind. To address these limitations, researchers have identified that hybrid systems that combine multiple power sources may be more effective than a battery-only PV/wind system. These hybrid systems use dynamic energy dispatch to optimize the overall cost and operation of the microgrid. Energy management systems are used to achieve this dynamic energy dispatch, which involves load profiling and intelligent decision-making on energy dispatch. While numerous energy management systems have been implemented worldwide to optimize microgrids, less work has been done in Africa and South Africa, specifically on the demand side energy management system. This paper focuses on a hybrid PV-driven battery and fuel cell backup system, with the initial aim of dimensioning the PV/battery and fuel cell. The work then shifted towards developing an energy management system. The proposed system was a low-power provision in a +48 VDC format, which covered lighting and powered computing, entertainment, and communication modules. Seven households were selected with a total energy consumption of 8640 W/h daily. The results compared the efficiency of microgrid with demand-side management versus without demand-side management, as well as the effect of load scheduling. The simulation was conducted using MATLAB/Simulink.
With the improved network performance and efficiency, 5G has been a very appealing alternative for various applications and devices, including but not limited to Industrial IoT (IIoT) applications. However, since IIoT...
详细信息
The process of choosing and narrowing down universities for graduate program admission plays a crucial role in the overall application procedure. This project aims to examine the uses of machine learning models to pre...
详细信息
The process of choosing and narrowing down universities for graduate program admission plays a crucial role in the overall application procedure. This project aims to examine the uses of machine learning models to predict the likelihood of a student being accepted into a master's program. Hence this will provide students with an early indication of their admission prospects. In the past, models were constructed using different algorithms such as random forest, multiple linear regression and k-nearest neighbor. Results have shown that logistic regression out performs these algorithms. The admission of students into educational institutions is a critical issue, and this study addresses the application of machine learning algorithms to predict the chance of admission of students into master's programs. These models include SVM, Gaussian Naive Bayes, and Logistic Regression, with experiments demonstrating that the Logistic Regression model surpasses the others. This will give students a better understanding of their admission prospects in advance.
Customer segmentation plays a crucial role in business development by providing deep insights into customer behaviors and preferences, facilitating effective and personalized marketing strategies. However, traditional...
Customer segmentation plays a crucial role in business development by providing deep insights into customer behaviors and preferences, facilitating effective and personalized marketing strategies. However, traditional segmentation methods often fall short of accurately analyzing large and complex datasets. In recent years, the application of machine learning techniques has transformed the field of customer segmentation, enabling automated analysis, enhanced accuracy, and the discovery of intricate patterns and key trends. This research paper provides a detailed review of the application of K-Means clustering in credit card companies for effective customer segmentation. It explores the benefits, challenges, and practical considerations of utilizing K-Means clustering, along with real-world case studies highlighting trends and potential advancements for using K-Means clustering in credit card customer segmentation.
In the rapidly evolving landscape of healthcare research, the demand for medical data has become indispensably vital. Additionally, incentivizing owners for sharing their health data while maintaining privacy and secu...
详细信息
In this research, the photocarrier transmission mechanism and the effect of the doping concentration of n-and p-strip regions in interdigitated back contact silicon heterojunction (IBC-SHJ) cell efficiency have been s...
In this research, the photocarrier transmission mechanism and the effect of the doping concentration of n-and p-strip regions in interdigitated back contact silicon heterojunction (IBC-SHJ) cell efficiency have been studied. In this regards, short-circuit current density, open-circuit voltage, fill factor and cell efficiency values have been evaluated using J-V curves for different conditions. The doping concentration of the n- and p-strip regions have been changed using trial-and-error method to achieve improved efficiency in IBC-SHJ solar cell. The improved IBC-SHJ have no extra ARCs and more structural periodicity. Thus, a simple structure with improved conversion efficiency is proposed. The results have been shown that the n- and p-strip doping concentration were the most effective parameters on efficiency improvement. According to the results, the best doping concentration of Emitter and BSF regions to achieve improved efficiency is equal to $2 \times 10^{19} \mathrm{~cm}^{-3}$ and $4.3 \times 10^{18} \mathrm{~cm}^{-3}$ respectively.
Social media has revolutionized communication and content publishing. Anonymity and mobility afforded by social media is also an effective medium for dissemination of hate messages, which can lead to undesired outcome...
Social media has revolutionized communication and content publishing. Anonymity and mobility afforded by social media is also an effective medium for dissemination of hate messages, which can lead to undesired outcomes in the society. Prevalence of hate speech is a global problem. This paper is intended to highlight the mechanisms to address the same in the Indian context. This work aims to provide a concise overview of the various machine learning, deep learning, and transformer methods to detect hate speech in social media text written in Indian languages, as detailed in recent research literature. Besides, a discussion some of the available monolingual and code-mixed datasets to train such classifiers and the commonly adopted strategies to address the data scarcity in Indian languages is also included.
The amount of data loss on corporate servers in cloud environments has increased significantly. In the cloud, there are many security compromises and account hijackings, resulting in severe vulnerabilities for the ser...
详细信息
暂无评论