Network of Tensor Time Series

TL;DR

This paper introduces a novel model combining tensor graph convolution and tensor recurrent networks to effectively capture explicit and implicit relationships in co-evolving time series data across various applications.

Contribution

It proposes a new network architecture that generalizes GCNs to tensor graphs and incorporates tensor decomposition for modeling complex temporal dynamics.

Findings

Outperforms existing methods on five real-world datasets

Effectively captures both explicit and implicit relationships

Demonstrates robustness across diverse applications

Abstract

Co-evolving time series appears in a multitude of applications such as environmental monitoring, financial analysis, and smart transportation. This paper aims to address the following challenges, including (C1) how to incorporate explicit relationship networks of the time series; (C2) how to model the implicit relationship of the temporal dynamics. We propose a novel model called Network of Tensor Time Series, which is comprised of two modules, including Tensor Graph Convolutional Network (TGCN) and Tensor Recurrent Neural Network (TRNN). TGCN tackles the first challenge by generalizing Graph Convolutional Network (GCN) for flat graphs to tensor graphs, which captures the synergy between multiple graphs associated with the tensors. TRNN leverages tensor decomposition to model the implicit relationships among co-evolving time series. The experimental results on five real-world datasets demonstrate the efficacy of the proposed method.