您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报(工学版) ›› 2012, Vol. 42 ›› Issue (1): 1-5.

• 机器学习与数据挖掘 •    下一篇

一种基于网络资本评价的关键核挖掘方法

黄金才,成清,刘彦君,乔士东   

  1. 国防科学技术大学信息系统工程重点实验室, 湖南 长沙 410073
  • 收稿日期:2011-04-15 出版日期:2012-02-20 发布日期:2011-04-15
  • 作者简介:黄金才(1973- ),男,安徽阜阳人,教授,博士,主要研究方向为机器学习. Email:huangjincai@gmail.com
  • 基金资助:

    国家自然科学基金资助项目(70771109, 71001105)

A method for critical core mining based on network capital evaluation

HUANG Jin-cai, CHENG Qing, LIU Yan-jun, QIAO Shi-dong   

  1. Science and Technology on Information System Engineering Laboratory, National University of
     Defense Technology,Changsha 410073, China
  • Received:2011-04-15 Online:2012-02-20 Published:2011-04-15

摘要:

为挖掘出网络中的多个关键点,定义了网络资本和关键核。根据网络特点,确定影响函数和衰减函数来计算网络的网络资本值,以网络资本的变化来反应网络性能变化。 根据投入产出分析的思想,提出了完全相关模型对网络的关键核进行挖掘,如果节点(集)动态删除后使网络性能下降到一定程度,则节点(集)为关键核。实验结果表明:当网络性能下降到原来的30%时,动态失效方法比按序失效方法挖掘关键核的点数少。本方法在对网络关键核的挖掘上更加准确和有效。

关键词: 网络资本, 关键核, 完全相关模型, 动态失效

Abstract:

To solve the problem of multiple critical nodes mining in the network, the definitions of network capital and critical core were  proposed. In addition,  the influence function and attenuation function were used to calculate network capital, and  the change of network capital was used  to characterize the change of network performance. Inspired by the inputoutput analysis, a complete correlated mode was proposed to mine the critical core, which was the node(set) and whose removal with its links decreased the network performance to a certain degree. Experimental results showed  that the dynamic failure method could mine the critical core with fewer nodes than the ranked failure method when the network performance decreased to 30%.This method was more accurate and efficient in mining the critical core.

Key words: network capital, critical core, complete correlated model, dynamic failure

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!