Time2Graph+: Bridging Time Series and Graph Representation Learning via Multiple Attentions

Abstract

Time series modeling has attracted great research interests in the last decades. Among the literature, shapelet-based models aim to extract representative subsequences, and could offer explanatory insights in the downstream tasks. But most of those works ignore the seasonal effects on the subsequences, as well as the evolutionary characteristics of shapelets. In order to capture the shapelet dynamics and evolutions, in this paper, we propose a novel framework of bridging time series representation learning and graph modeling, with two different implementations. We first formulate the process of extracting time-aware shapelets by directly adding time-level attentions, then introduce the key idea of transforming time series data into shapelet evolution graphs, to model the shapelet evolutionary patterns. A straightforward solution is to enumerate all possible shapelet transitions among adjacent time series segments, and apply a random-walk-based graph embedding algorithm to learn time series representations (Time2Graph). We further extend Time2Graph by adopting graph attention mechanism to refine the procedure of modeling shapelet evolutions, namely Time2Graph+. Specifically, we transform each time series data into a unique unweighted shapelet graph, and use GAT to automatically capture the correlations between shapelets. Experimental results on three real-world datasets show the significant improvements of Time2Graph+ over Time2Graph and 17 baseline methods, and observational analysis demonstrates the effectiveness and interpretability brought by both time-level and graph-level attentions. Furthermore, the success of online deployment of Time2Graph+ model in State Grid of China validates the whole framework in the real-world application. Codes and documentations are available at https://github.com/petecheng/Time2GraphPlus.