Temporal Graph Neural Networks for Irregular Data

Joel Oskarsson,Per Sidén,Fredrik Lindsten

DOI: https://doi.org/10.48550/arXiv.2302.08415

2023-02-17

Abstract:This paper proposes a temporal graph neural network model for forecasting of graph-structured irregularly observed time series. Our TGNN4I model is designed to handle both irregular time steps and partial observations of the graph. This is achieved by introducing a time-continuous latent state in each node, following a linear Ordinary Differential Equation (ODE) defined by the output of a Gated Recurrent Unit (GRU). The ODE has an explicit solution as a combination of exponential decay and periodic dynamics. Observations in the graph neighborhood are taken into account by integrating graph neural network layers in both the GRU state update and predictive model. The time-continuous dynamics additionally enable the model to make predictions at arbitrary time steps. We propose a loss function that leverages this and allows for training the model for forecasting over different time horizons. Experiments on simulated data and real-world data from traffic and climate modeling validate the usefulness of both the graph structure and time-continuous dynamics in settings with irregular observations.

Machine Learning,Social and Information Networks

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the time prediction problem of handling irregular observational data in graph - structured time series. Specifically, the paper focuses on how to conduct effective graph - structured time - series prediction in the case of irregular sampling in time and some nodes not being observed at certain time points. Such problems are very common in fields such as traffic flow prediction and climate modeling. However, traditional graph neural networks (GNN) and time - series prediction methods often assume that data is sampled at a fixed frequency and has no missing values, which is inconsistent with the actual situation.

To address these challenges, the authors propose a model named **TGNN4I**. This model deals with irregular time intervals and partial observations by introducing a time - continuous hidden state for each node. This hidden state follows a linear ordinary differential equation (ODE) defined by the output of the gated recurrent unit (GRU) and can be represented as a combination of exponential decay and periodic dynamics through an explicit solution. In addition, the model also integrates graph neural network layers so as to consider the observed values in the graph neighborhood in the hidden - state update and prediction model. This design enables the model to make predictions at any time point and adapt to prediction tasks within different time ranges.

In summary, the main contribution of this paper lies in proposing a new type of temporal graph neural network model that can effectively handle irregular observational data in graph - structured time series.