Graph Neural Flows for Unveiling Systemic Interactions Among Irregularly Sampled Time Series

TL;DR

This paper introduces a graph neural flow model that captures systemic interactions among irregularly sampled time series using a directed acyclic graph and continuous-time modeling, improving prediction accuracy.

Contribution

It presents a novel graph neural flow approach that models conditional dependencies among time series components and learns these dependencies jointly with continuous-time dynamics.

Findings

Significant improvements in time series classification accuracy.

Enhanced forecasting performance over baseline methods.

Effective modeling of causal dependencies among system components.

Abstract

Interacting systems are prevalent in nature. It is challenging to accurately predict the dynamics of the system if its constituent components are analyzed independently. We develop a graph-based model that unveils the systemic interactions of time series observed at irregular time points, by using a directed acyclic graph to model the conditional dependencies (a form of causal notation) of the system components and learning this graph in tandem with a continuous-time model that parameterizes the solution curves of ordinary differential equations (ODEs). Our technique, a graph neural flow, leads to substantial enhancements over non-graph-based methods, as well as graph-based methods without the modeling of conditional dependencies. We validate our approach on several tasks, including time series classification and forecasting, to demonstrate its efficacy.