Scalability and information personal privacy are vital for training and deploying large-scale deep learning models.Federated learning trains models on exclusive information by aggregating weights from various devices ...Scalability and information personal privacy are vital for training and deploying large-scale deep learning models.Federated learning trains models on exclusive information by aggregating weights from various devices and taking advantage of the device-agnostic environment of web browsers.Nevertheless,relying on a main central server for internet browser-based federated systems can prohibit scalability and interfere with the training process as a result of growing client numbers.Additionally,information relating to the training dataset can possibly be extracted from the distributed weights,potentially reducing the privacy of the local data used for training.In this research paper,we aim to investigate the challenges of scalability and data privacy to increase the efficiency of distributed training models.As a result,we propose a web-federated learning exchange(WebFLex)framework,which intends to improve the decentralization of the federated learning process.WebFLex is additionally developed to secure distributed and scalable federated learning systems that operate in web browsers across heterogeneous devices.Furthermore,WebFLex utilizes peer-to-peer interactions and secure weight exchanges utilizing browser-to-browser web real-time communication(WebRTC),efficiently preventing the need for a main central server.WebFLex has actually been measured in various setups using the MNIST dataset.Experimental results show WebFLex’s ability to improve the scalability of federated learning systems,allowing a smooth increase in the number of participating devices without central data aggregation.In addition,WebFLex can maintain a durable federated learning procedure even when faced with device disconnections and network variability.Additionally,it improves data privacy by utilizing artificial noise,which accomplishes an appropriate balance between accuracy and privacy preservation.展开更多
一引言
90年代初,一种新型的计算机网络应用技术--电子数据交换EDI(Electronic Data Interchange)以其特有的简洁、高效、安全和迅捷特性引起世界各国的高度重视,被认为是提高工作效率、服务质量和企业竞争能力的强有力的手段[1].EDI旨...一引言
90年代初,一种新型的计算机网络应用技术--电子数据交换EDI(Electronic Data Interchange)以其特有的简洁、高效、安全和迅捷特性引起世界各国的高度重视,被认为是提高工作效率、服务质量和企业竞争能力的强有力的手段[1].EDI旨在实现表单传送的电子化,所以有人称EDI为无纸化贸易.使用电子表单的同时仍然需要纸张表单辅助,只是纸张表单从先前的主要或唯一的地位,下降到次要和辅助的地位.也就是说,EDI最重要的意义不在于节约纸张,而在于其快速、避免重复劳动、提高效率、节约成本等方面,因此EDI技术的实质是强调快速传输(比如从邮寄的几天变成几分钟甚至实时)、节约劳动(不必反复打印和录入表单),从而提高效率和节约成本.展开更多
确保无线电波的质量和安全播出是广播电视行业的基本要求,依赖无线发射台站的高效管理。传统的基于客户端的台站管理系统因其高成本、低效率等缺点逐渐不能满足当前的需求。因此,提出基于WebGIS(Web Geographic Information System)的...确保无线电波的质量和安全播出是广播电视行业的基本要求,依赖无线发射台站的高效管理。传统的基于客户端的台站管理系统因其高成本、低效率等缺点逐渐不能满足当前的需求。因此,提出基于WebGIS(Web Geographic Information System)的广播电视无线发射台站管理系统,旨在提升广播电视无线发射台站的管理能力。首先阐述系统特点,其次设计系统的总体结构,分析系统的主要功能模块,最后研究各功能模块的实现。展开更多
讨论了基于M atlab W eb Server的M atlab网络应用开发原理,介绍了M atlab W eb程序处理的一般流程和相关配置文件的详细配置方法,并给出M atlab W eb开发中的两个关键问题:通过输入模块从HTML页面获取输入参数和通过输出模块生成包括...讨论了基于M atlab W eb Server的M atlab网络应用开发原理,介绍了M atlab W eb程序处理的一般流程和相关配置文件的详细配置方法,并给出M atlab W eb开发中的两个关键问题:通过输入模块从HTML页面获取输入参数和通过输出模块生成包括输出数据和图片的HTML文件.利用M atlab W eb Server环境实现了远程控制实验室的控制效果仿真,并以二维图形的输出形式显示仿真结果,为网上控制实验室的建立提供了控制参数选择以及试验结果验证参照.本远程数据处理方法可推广应用到不同的远程数据处理领域,具有很高的推广价值.展开更多
在介绍Matlab Web Server的工作原理的基础上,结合通信原理远程仿真程序的开发实例,详细说明了基于Matlab Web Server的远程仿真系统的关键方法和技巧。仿真系统用户通过Web浏览器在远程输入数据,提交给MATLAB Web Server上的MATLAB运行...在介绍Matlab Web Server的工作原理的基础上,结合通信原理远程仿真程序的开发实例,详细说明了基于Matlab Web Server的远程仿真系统的关键方法和技巧。仿真系统用户通过Web浏览器在远程输入数据,提交给MATLAB Web Server上的MATLAB运行,最后将计算结果和图形直观地显示在浏览器上。展开更多
基金This work has been funded by King Saud University,Riyadh,Saudi Arabia,through Researchers Supporting Project Number(RSPD2024R857).
文摘Scalability and information personal privacy are vital for training and deploying large-scale deep learning models.Federated learning trains models on exclusive information by aggregating weights from various devices and taking advantage of the device-agnostic environment of web browsers.Nevertheless,relying on a main central server for internet browser-based federated systems can prohibit scalability and interfere with the training process as a result of growing client numbers.Additionally,information relating to the training dataset can possibly be extracted from the distributed weights,potentially reducing the privacy of the local data used for training.In this research paper,we aim to investigate the challenges of scalability and data privacy to increase the efficiency of distributed training models.As a result,we propose a web-federated learning exchange(WebFLex)framework,which intends to improve the decentralization of the federated learning process.WebFLex is additionally developed to secure distributed and scalable federated learning systems that operate in web browsers across heterogeneous devices.Furthermore,WebFLex utilizes peer-to-peer interactions and secure weight exchanges utilizing browser-to-browser web real-time communication(WebRTC),efficiently preventing the need for a main central server.WebFLex has actually been measured in various setups using the MNIST dataset.Experimental results show WebFLex’s ability to improve the scalability of federated learning systems,allowing a smooth increase in the number of participating devices without central data aggregation.In addition,WebFLex can maintain a durable federated learning procedure even when faced with device disconnections and network variability.Additionally,it improves data privacy by utilizing artificial noise,which accomplishes an appropriate balance between accuracy and privacy preservation.
文摘一引言
90年代初,一种新型的计算机网络应用技术--电子数据交换EDI(Electronic Data Interchange)以其特有的简洁、高效、安全和迅捷特性引起世界各国的高度重视,被认为是提高工作效率、服务质量和企业竞争能力的强有力的手段[1].EDI旨在实现表单传送的电子化,所以有人称EDI为无纸化贸易.使用电子表单的同时仍然需要纸张表单辅助,只是纸张表单从先前的主要或唯一的地位,下降到次要和辅助的地位.也就是说,EDI最重要的意义不在于节约纸张,而在于其快速、避免重复劳动、提高效率、节约成本等方面,因此EDI技术的实质是强调快速传输(比如从邮寄的几天变成几分钟甚至实时)、节约劳动(不必反复打印和录入表单),从而提高效率和节约成本.
文摘确保无线电波的质量和安全播出是广播电视行业的基本要求,依赖无线发射台站的高效管理。传统的基于客户端的台站管理系统因其高成本、低效率等缺点逐渐不能满足当前的需求。因此,提出基于WebGIS(Web Geographic Information System)的广播电视无线发射台站管理系统,旨在提升广播电视无线发射台站的管理能力。首先阐述系统特点,其次设计系统的总体结构,分析系统的主要功能模块,最后研究各功能模块的实现。
文摘讨论了基于M atlab W eb Server的M atlab网络应用开发原理,介绍了M atlab W eb程序处理的一般流程和相关配置文件的详细配置方法,并给出M atlab W eb开发中的两个关键问题:通过输入模块从HTML页面获取输入参数和通过输出模块生成包括输出数据和图片的HTML文件.利用M atlab W eb Server环境实现了远程控制实验室的控制效果仿真,并以二维图形的输出形式显示仿真结果,为网上控制实验室的建立提供了控制参数选择以及试验结果验证参照.本远程数据处理方法可推广应用到不同的远程数据处理领域,具有很高的推广价值.
文摘在介绍Matlab Web Server的工作原理的基础上,结合通信原理远程仿真程序的开发实例,详细说明了基于Matlab Web Server的远程仿真系统的关键方法和技巧。仿真系统用户通过Web浏览器在远程输入数据,提交给MATLAB Web Server上的MATLAB运行,最后将计算结果和图形直观地显示在浏览器上。
