首页> 外文学位 >Robust linear separation of multiple finite sets.
【24h】

Robust linear separation of multiple finite sets.

机译:多个有限集的鲁棒线性分离。

获取原文
获取原文并翻译 | 示例

摘要

Given K disjoint finite sets AkK k=1 in Euclidean n-space, a general problem with numerous applications is to find K simple nontrivial functions fk(x) which separate the sets {lcub} Ak{rcub} in the sense that fk( x) ≤ fi(x) for all a in Ak and i ≠ k, k = 1,...,K. Typically one seeks linear functions fk(x). When such linear functions exist we say that the sets {lcub}Ak{rcub} are simply separable. If the sets are simply separable, there are generally many functions that separate. In this case we seek a 'best' separator in the sense that a measure of Euclidean distance between the separator and the sets is as large as possible. That separator is referred as a robust separator. If the sets are not simply separable, we seek a function that separates the sets pairwise. In this case we are also looking for a robust pairwise separator. In general, the K sets AkK k=1 can always be separated with the piecewise linear function obtained by the Voronoi Partition defined for the points in ∪Kk=1Ak . This function provides the robust separator by definition.; Three distinct robust approaches are proposed in this dissertation: Simple Separation of multiple sets, Pairwise Separation, and separation via the Voronoi Partition. The Simple Separation procedure can be applied to sets that are simply separable. The second approach, Pairwise Separation, is used when the sets are not simply separable, but are separable pairwise. Finally, the Voronoi Partition can be applied to any collection of sets.; To test the efficiency of our methodology, two types of computational experiments were performed: one based on randomly generated data, and the other based on established databases. In both cases, we show that our methods are efficient and accurate.
机译:给定欧几里得n空间中的K个不相交的有限集AkK k = 1,许多应用程序中的一个普遍问题是找到K个简单的非平凡函数fk(x),它们将集{lcub} Ak {rcub}分离为fk(x )≤fi(x)对于Ak中的所有a且i≠k,k = 1,...,K。通常,人们寻求线性函数fk(x)。当存在这样的线性函数时,我们说{lcub} Ak {rcub}集是可分离的。如果这些集合可以简单地分开,那么通常会有许多分开的功能。在这种情况下,我们在某种意义上寻求“最佳”分隔符,因为分隔符与集合之间的欧几里德距离的量度应尽可能大。该分离器称为鲁棒分离器。如果集合不是简单可分离的,我们将寻找一个将集合成对分离的函数。在这种情况下,我们也在寻找一种健壮的成对分隔符。通常,K个集合AkK k = 1总是可以用分段线性函数分开,该分段线性函数是由Voronoi分区为∪Kk= 1Ak中的点定义的。该函数根据定义提供了强大的分隔符。本文提出了三种不同的鲁棒方法:多组简单分离,成对分离和通过Voronoi分区分离。简单分离过程可以应用于简单可分离的集合。第二种方法是成对分离,当集合不能简单地分离而是成对分离时。最后,Voronoi分区可以应用于任何集合集合。为了测试我们方法的效率,执行了两种类型的计算实验:一种基于随机生成的数据,另一种基于已建立的数据库。在这两种情况下,我们都表明我们的方法是有效且准确的。

著录项

  • 作者

    Yeganova, Liana Edward.;

  • 作者单位

    The George Washington University.;

  • 授予单位 The George Washington University.;
  • 学科 Operations Research.
  • 学位 D.Sc.
  • 年度 2001
  • 页码 85 p.
  • 总页数 85
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 运筹学;
  • 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