基于层次随机图模型的脑网络链路预测

针对一般链路预测算法在具有层次结构的脑网络中计算效率低且复杂度高的问题，提出了一种基于最大似然估计的层次随机图模型。该算法首先利用脑网络数据建立层次随机图，然后通过改进的马尔科夫蒙特卡罗算法采样树状图空间，最后计算脑网络边的平均连接概率，且通过评价指标对算法进行评价。实验结果表明，利用该算法对脑网络和3种不同的层次结构网络进行链路预测比较, 脑网络的预测结果最好。此外，所提出的算法较之传统的基于相似性的算法，该算法效果明显，且具有理想的计算复杂度。

Link prediction of brain networks based on hierarchical random graph model

Focusing on the problem of poor efficiency and high complexity in general link prediction algorithms applied in brain networks, this paper proposed a hierarchical random graph model based on maximum likelihood estimation. Firstly, this algorithm used brain networks data to create the hierarchical random graph model. Then, it sampled the space of all possible dendrograms using a improved Markov chain Monte Carlo algorithm. Finally, it calculated the average connection probability of brain network edges, and it also evaluated by the evaluation index. Experimental results show that the algorithm exhibits best result in brain network between different hierarchical networks. In addition, it obtains the good effect and reasonable computing complexity compared with the traditional algorithms based on similarity.