复杂网络幂律度分布和层次聚集函数标度指数的一种新的估计方法

A New Method to Estimate Scaling Exponents of Power-Law Degree Distribution and Hierarchical Clustering Function for Complex Networks

提出一种估计复杂网络幂律度分布和层次聚集函数标度指数的新方法,并给出求解这些指数的数值算法.该方法可以克服目前网络研究中采用的图形线性拟合估计方法存在的偏差和不准确等不足之处.此外,通过对10个CNN网络进行KS检验统计量的比较,证实该方法比图形方法具有更好的拟合效果.

A new method and corresponding numerical procedure were introduced to estimate scaling exponents of power-law degree distibution and hierarchical clustering function for complex networks. This method could overcome the biased and inaccurate faults of graphical linear fitting methods commonly used in current network research. Furthermore, it has been verified to have higher goodnessof-fit than graphical methods by comparing the KS test statistics for 10 CNN networks.