An attentive Copula-based spatio-temporal graph model for multivariate time-series forecasting

Xihe Qiu, Jiahui Qian, Haoyu Wang, Xiaoyu Tan, Yaochu Jin

Abstract: Highlights•We use an adjacency matrix with a Gaussian kernel for a spatially efficient graph.•It’s a graph learning framework for joint iterative learning of discrete structures.•Copula enhances spatio-temporal graphs with improved correlation and representation.