Discovering Latent Dependence of Large Volatility Events

Learning the latent dependence from time-series data is of great interest in network analysis, while most existing methods usually ignore the large volatility underlying the time series. In this paper, we develop a probabilistic model that leverages mutually exciting Hawkes processes with network graph models to infer the latent dependence of large volatility events. Specifically, the Hawkes process is used to describe the influence of volatility and the network graph models the dependence. Extensive empirical evaluations are conducted on both synthetic and two real-world datasets. The results demonstrate the effectiveness of our proposed model in learning the latent spatio-temporal dependence.