arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1902.00035physics.soc-phcond-mat.stat-mechcs.SI
收藏

复杂网络中缺失环节的可预测性

Predictability of missing links in complex networks

Guillermo García-Pérez, Roya Aliakbarisani, Abdorasoul Ghasemi, M. Ángeles Serrano

摘要:

预测真实网络中的缺失链路是网络科学中的一个重要问题,人们在这方面做了大量的工作,因此在文献中出现了大量的链路预测方法。在这项工作中,我们采取了不同的观点对这个问题,并研究的理论局限性的可预测性的缺失环节。特别是,我们假设有一个不可约的不确定性,在真实网络的链接预测作为其形成过程的随机性的结果。通过考虑由著名的网络模型定义的合奏,我们证明了分析,即使是最好的可能的链接预测方法的合奏,给出的合奏连接概率的排名,产生有限的精度。这一结果表明,在实际复杂网络中,链接的可预测性在理论上是有限的。最后,我们表明,连接概率推断拟合网络模型真实的网络允许估计一个上限的可预测性丢失的链接,我们进一步提出了一种方法来近似这样的界限从不完整的实例真实世界的网络。

英文摘要:

Predicting missing links in real networks is an important problem in network science to which considerable efforts have been devoted, giving as a result a vast plethora of link prediction methods in the literature. In this work, we take a different point of view on the problem and study the theoretical limitations to the predictability of missing links. In particular, we hypothesise that there is an irreducible uncertainty in link prediction on real networks as a consequence of the random nature of their formation process. By considering ensembles defined by well-known network models, we prove analytically that even the best possible link prediction method for an ensemble, given by the ranking of the ensemble connection probabilities, yields a limited precision. This result suggests a theoretical limitation to the predictability of links in real complex networks. Finally, we show that connection probabilities inferred by fitting network models to real networks allow to estimate an upper-bound to the predictability of missing links, and we further propose a method to approximate such bound from incomplete instances of real-world networks.

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。

↑