• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Graph coloring and clustering algorithms for science and engineering applications.
【24h】

Graph coloring and clustering algorithms for science and engineering applications.

机译：用于科学和工程应用的图形着色和聚类算法。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In this dissertation, efficient algorithms are proposed for two important combinatorial problems in science and engineering applications: parallel graph coloring and subspace clustering. For parallel distance-1 and distance-2 coloring problems, a framework that unifies several existing techniques for creating or facilitating concurrency are presented for distributed-memory computers. Extensions for similar graph and hypergraph coloring problems are also described. Through extensive set of experiments, the algorithms are shown to be efficient and scalable. In addition, new algorithms are proposed for particular instances of subspace clustering problem. A generalized cluster pattern model is introduced to discover complex relationships in large datasets and a search algorithm that utilizes established techniques from related combinatorial problems is designed to solve this problem. Also, biclustering, a special case of subspace clustering, is considered and a new biclustering algorithm is proposed to identify co-regulated genes in large gene expression datasets. The algorithm is the first one to utilize a well known statistical correlation measure in biclustering context. In order to extract useful correlation information from multiple gene expression datasets, a novel method is introduced as well. Both subspace clustering and biclustering algorithms are shown to perform well at identifying targeted relationships in large datasets. The biclustering algorithm is also applied on real gene expression datasets and returned promising clustering results.
机译：本文针对科学和工程应用中的两个重要组合问题，提出了高效的算法：并行图着色和子空间聚类。对于并行的距离1和距离2着色问题，为分布式内存计算机提供了一种框架，该框架统一了用于创建或促进并发的几种现有技术。还介绍了类似图形和超图着色问题的扩展。通过广泛的实验，该算法被证明是高效且可扩展的。另外，针对子空间聚类问题的特定实例提出了新算法。引入了一种通用的集群模式模型来发现大型数据集中的复杂关系，并设计了一种利用相关组合问题中已建立的技术的搜索算法来解决该问题。此外，考虑了双聚类，这是子空间聚类的一种特殊情况，并提出了一种新的双聚类算法来识别大型基因表达数据集中的共同调控基因。该算法是第一种在双聚类环境中利用众所周知的统计相关性度量的算法。为了从多个基因表达数据集中提取有用的相关信息，还引入了一种新的方法。子空间聚类和双聚类算法均显示出在识别大型数据集中的目标关系时表现良好。双聚类算法也应用于真实基因表达数据集，并返回有希望的聚类结果。

著录项

  • 作者

    Bozdag, Doruk.;

  • 作者单位

    The Ohio State University.;

  • 授予单位 The Ohio State University.;
  • 学科 Engineering Electronics and Electrical.
  • 学位 Ph.D.
  • 年度 2009
  • 页码 179 p.
  • 总页数 179
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 无线电电子学、电信技术;
  • 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