arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1405.5756physics.soc-phcs.SI
收藏

采用成本效益高的抽样方法

Towards Cost-efficient Sampling Methods

Luo Peng, Li Yongli, Wu Chong

摘要:

抽样方法在复杂网络领域,特别是统计物理领域一直受到广泛的关注。基于一小部分节点度高的顶点可以拥有网络的大部分结构信息的观点,提出了两种新的采样方法。这两种采样方法在对度较高的节点进行采样时是有效的。第一种新的抽样方法是在分层随机抽样方法的基础上改进的,通过对节点的度分布进行分类,选择概率较高的度节点。第二种采样方法改进了现有的滚雪球采样方法,使其能够在每个采样步骤中选择性地对目标节点进行采样。此外,这两种采样方法不仅对节点进行采样,而且对与这些节点直接相连的边进行采样。为了证明这两种方法的有效性和准确性,我们在三种常用的模拟网络(无标度网络、随机网络、小世界网络)和两种真实网络中将它们与现有的抽样方法进行了比较。实验结果表明,本文提出的两种采样方法在采样代价和获取真实网络结构特征方面均优于现有的采样方法。

英文摘要:

The sampling method has been paid much attention in the field of complex network in general and statistical physics in particular. This paper presents two new sampling methods based on the perspective that a small part of vertices with high node degree can possess the most structure information of a network. The two proposed sampling methods are efficient in sampling the nodes with high degree. The first new sampling method is improved on the basis of the stratified random sampling method and selects the high degree nodes with higher probability by classifying the nodes according to their degree distribution. The second sampling method improves the existing snowball sampling method so that it enables to sample the targeted nodes selectively in every sampling step. Besides, the two proposed sampling methods not only sample the nodes but also pick the edges directly connected to these nodes. In order to demonstrate the two methods' availability and accuracy, we compare them with the existing sampling methods in three commonly used simulation networks that are scale-free network, random network, small-world network, and two real networks. The experimental results show that the two proposed sampling methods perform much better than the compared existing sampling methods in terms of sampling cost and obtaining the true network structural characteristics.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑