This paper studies the change point detection problem in time series of networks, with the Separable Temporal Exponential-family Random Graph Model (STERGM). We consider a sequence of networks generated from a piecewise constant distribution that is altered at unknown change points in time. Detection of the change points can identify the discrepancies in the underlying data generating processes and facilitate downstream dynamic network analysis tasks. Moreover, the STERGM that focuses on network statistics is a flexible model to fit dynamic networks with both dyadic and temporal dependence. We propose a new estimator derived from the Alternating Direction Method of Multipliers (ADMM) and the Group Fused Lasso to simultaneously detect multiple time points, where the parameters of STERGM have changed. We also provide Bayesian information criterion for model selection to assist the detection. Our experiments show good performance of the proposed method on both simulated and real data. Lastly, we develop an R package CPDstergm to implement our method.

翻译：本文研究网络时间序列中的变点检测问题，采用可分离时间指数族随机图模型（STERGM）进行分析。我们考虑由分段常数分布生成的网络序列，该分布在未知时间点发生突变。检测这些变点能够识别底层数据生成过程中的差异，并促进下游动态网络分析任务。此外，专注于网络统计量的STERGM是一种灵活模型，可同时拟合具有二元依赖性和时间依赖性的动态网络。我们提出一种基于交替方向乘子法（ADMM）和分组融合套索的新估计量，用以同时检测STERGM参数发生变化的多个时间点。同时，我们提供贝叶斯信息准则进行模型选择以辅助检测。实验结果表明，所提方法在模拟数据和真实数据上均表现良好。最后，我们开发了R语言包CPDstergm以实现该方法。