Time-varying Interaction Graph ODE for Dynamic Graph Representation Learning

2026-04-27

AI Summary

Existing graph neural ODEs employ a uniform message-passing mechanism on dynamic graphs, which struggles to capture the diversity and time-varying nature of node interactions. This work proposes TI-ODE, the first model to incorporate a time-varying interaction basis function mechanism: it decomposes the evolution function into a set of learnable interaction basis functions and dynamically combines them via time-adaptive, learnable weights, enabling flexible modeling of continuous-time dynamic graphs. This approach substantially enhances model expressiveness, interpretability, and robustness, achieving state-of-the-art performance on node property prediction across six dynamic graph benchmarks. Furthermore, experiments on a COVID-19 dataset demonstrate its strong generalization capability and interpretability.

Abstract

Graph neural Ordinary Differential Equations (ODE) combine neural ODE with the message passing mechanism of Graph Neural Networks (GNN), providing a continuous-time modeling method for graph representation learning. However, in dynamic graph scenarios, existing graph neural ODEs typically employ a unified message passing mechanism, assuming that inter-node interactions share the same message passing function at any time, which makes it challenging to capture the diversity and time-varying nature of inter-node interaction patterns. To address this, we propose Time-varying Interaction Graph Ordinary Differential Equations (TI-ODE). The core idea of TI-ODE is to decompose the evolution function of a graph ODE into a set of learnable interaction basis functions, where each basis function corresponds to a distinct type of inter-node interaction. These basis functions are dynamically combined through time-dependent learnable weights, enabling inter-node interaction patterns to adaptively evolve over time. Experimental results on six dynamic graph datasets demonstrate that TI-ODE consistently outperforms existing methods and achieves state-of-the-art performance on attribute prediction tasks, and experiments on the \textit{Covid} dataset further verify the interpretability and generalizability of our TI-ODE. Furthermore, we demonstrate both theoretically and empirically that TI-ODE exhibits superior robustness compared to models utilizing a unified message-passing mechanism.

Problem

• dynamic graph
• graph neural ODE
• time-varying interaction
• message passing
• representation learning

Innovation

• Time-varying Interaction
• Graph Neural ODE
• Dynamic Graph Representation Learning
• Learnable Basis Functions
• Adaptive Message Passing