首页> 外文期刊>IEEE Signal Processing Magazine >Distributed Gradient Methods for Convex Machine Learning Problems in Networks: Distributed Optimization
【24h】

Distributed Gradient Methods for Convex Machine Learning Problems in Networks: Distributed Optimization

机译:网络中凸机学习问题的分布式梯度方法:分布式优化

获取原文
获取原文并翻译 | 示例
           

摘要

This article provides an overview of distributed gradient methods for solving convex machine learning problems of the form minx & in; R-n (1/m) & sum;(m)(i = 1) f(i)(x) in a system consisting of m agents that are embedded in a communication network. Each agent i has a collection of data captured by its privately known objective function f(i)(x). The distributed algorithms considered here obey two simple rules: privately known agent functions f(i)(x) cannot be disclosed to any other agent in the network and every agent is aware of the local connectivity structure of the network, i.e., it knows its one-hop neighbors only. While obeying these two rules, the distributed algorithms that agents execute should find a solution to the overall system problem with the limited knowledge of the objective function and limited local communications. Given in this article is an overview of such algorithms that typically involve two update steps: a gradient step based on the agent local objective function and a mixing step that essentially diffuses relevant information from one to all other agents in the network.
机译:本文概述了用于求解MINX&IN的凸机学习问题的分布式梯度方法; R-N(1 / m)和总和;(m)(i = 1)f(i)(i)(x)由嵌入通信网络中的M代理组成。每个代理商我都有一个由其私人已知的目标函数f(i)(x)捕获的数据集合。这里考虑的分布式算法遵循了两个简单的规则:私人已知的代理函数f(i)(x)不能向网络中的任何其他代理公开,并且每个代理都知道网络的本地连接结构,即它知道它只有一个跳的邻居。在遵循这两个规则的同时,代理执行的分布式算法应该以有限的客观函数和本地通信有限的了解找到整体系统问题的解决方案。在本文中给出的是这种算法的概述,这些算法通常涉及两个更新步骤:基于代理局本地目标函数的梯度步骤和基本上扩散到网络中所有其他代理的相关信息的混合步骤。

著录项

  • 来源
    《IEEE Signal Processing Magazine》 |2020年第3期|92-101|共10页
  • 作者

    Nedic Angelia;

  • 作者单位

    Arizona State Univ Sch Elect Comp & Energy Engn Tempe AZ 85004 USA|Univ Illinois Champaign IL USA;

  • 收录信息
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号