arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2005.07373cs.DCcs.DS
收藏

$K$ -最近邻问题的高效分布式算法

Efficient Distributed Algorithms for the $K$-Nearest Neighbors Problem

Reza Fathi, Anisur Rahaman Molla, Gopal Pandurangan

摘要:

$K$ -最近邻是机器学习中的一个基本问题,有着广泛的应用。在这个问题中,给定一组带有标签的$n$数据点和一个查询点$p$,我们希望根据与查询最近的$K$点的标签为$p$分配一个标签。我们在{ \em $k$ -机器模型}中研究这个问题,(注意参数$k$代表$k$ -机器模型中的机器数量,并且与$K$ -最近点无关。)分布式大规模数据模型。在这个模型中,我们假设$n$点在$k$机器之间分布(以平衡的方式),目标是快速计算给定机器查询点的答案。我们的主要结果是一个简单的随机算法在$k$ -机器模型,运行在$O(\log K)$通信轮具有高概率的成功(无论机器的数量$k$和点的数量$n$)。该算法的消息复杂度小,只需要$O(k\log K)$消息。我们的边界本质上是基于比较的算法(仅使用元素之间的比较操作($\leq, \geq, =$)来区分它们之间的顺序的算法)的最佳可能。这是由于存在$Ω(\log n)$通信循环的下限,用于由Rodeh \cite{rodeh}找到在两个处理器之间均匀分布的$2n$元素的{ \em中值}。我们还实现了我们的算法,并表明它的性能良好的算法(在实践中使用),发送$K$最近的点从每台机器到一台机器,然后计算答案。

英文摘要:

The $K$-nearest neighbors is a basic problem in machine learning with numerous applications. In this problem, given a (training) set of $n$ data points with labels and a query point $p$, we want to assign a label to $p$ based on the labels of the $K$-nearest points to the query. We study this problem in the {\em $k$-machine model}, (Note that parameter $k$ stands for the number of machines in the $k$-machine model and is independent of $K$-nearest points.) a model for distributed large-scale data. In this model, we assume that the $n$ points are distributed (in a balanced fashion) among the $k$ machines and the goal is to quickly compute answer given a query point to a machine. Our main result is a simple randomized algorithm in the $k$-machine model that runs in $O(\log K)$ communication rounds with high probability success (regardless of the number of machines $k$ and the number of points $n$). The message complexity of the algorithm is small taking only $O(k\log K)$ messages. Our bounds are essentially the best possible for comparison-based algorithms (Algorithms that use only comparison operations ($\leq, \geq, =$) between elements to distinguish the ordering among them). This is due to the existence of a lower bound of $Ω(\log n)$ communication rounds for finding the {\em median} of $2n$ elements distributed evenly among two processors by Rodeh \cite{rodeh}. We also implemented our algorithm and show that it performs well compared to an algorithm (used in practice) that sends $K$ nearest points from each machine to a single machine which then computes the answer.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑