基于多尺度几何分析的复杂网络压缩策略  

Compression of Complex Networks Based on Multiscale Geometric Analysis

作　　者：王晓华[1] 杨新艳[1,2] 焦李成[1] 

机构地区：[1]西安电子科技大学智能信息处理研究所,西安710071 [2]苏州大学电子信息学院,苏州215021

出　　处：《电子与信息学报》2009年第4期968-972,共5页Journal of Electronics & Information Technology

基　　金：国家863计划项目(2006AA01Z107);国家部委科技项目(XADZ2008159)资助课题

摘　　要：自然界许多系统都具有多尺度效应,人们对系统的观察或测量也往往是在不同尺度上进行的。复杂网络可视为描述复杂系统的抽象模型,多尺度几何分析提供一种有效刻画复杂系统结构由粗到细的显微镜策略。该文从多尺度几何分析的角度对复杂网络压缩问题进行了有益的尝试,提出一种有效的网络压缩策略。策略分析及通过对不同拓扑类型的网络仿真表明基于多尺度几何分析的复杂网络压缩策略是可行的,能有效地帮助人们通过尽可能少的信息来分析、刻画大型复杂网络,实现网络数据及结构的稀疏表示。Many systems are characterized by multiscale features in nature, and the systems are frequently observed and measured in different scales. Complex networks can be regarded as the refined model for describing complex systems. The multiscale geometric analysis provides a powerful microscopy to observe the structures of complex systems. Consequently, complex networks can be analyses via multiscale geometric analysis. Keeping in mind this ides, a compression strategy based on multiscale geometric analysis is proposed for depicting the structure of complex networks with the small amount of information. The analysis and the experiments indicate that the proposed method is feasible, and performs as a sparsely representing for successfully understanding and compressing networks, which can be useful for analyzing networks with the small amount of information.

关 键 词：复杂网络 多尺度几何分析 网络压缩 稀疏表示 

分 类 号：TN711[电子电信—电路与系统] O231.5[理学—运筹学与控制论]