Evolving Graph Learning for Out-of-Distribution Generalization in Non-stationary Environments

TL;DR

This paper introduces EvoOOD, a framework that enhances out-of-distribution generalization in dynamic graph neural networks by modeling environment evolution and invariant pattern recognition, addressing challenges posed by non-stationary environments.

Contribution

It proposes a novel environment-aware invariant pattern recognition method combined with environment evolution modeling for dynamic graph OOD generalization.

Findings

EvoOOD outperforms existing methods on real-world and synthetic datasets.

The approach effectively captures environment evolution for better OOD prediction.

First study on dynamic graph OOD generalization from environment evolution perspective.

Abstract

Graph neural networks have shown remarkable success in exploiting the spatial and temporal patterns on dynamic graphs. However, existing GNNs exhibit poor generalization ability under distribution shifts, which is inevitable in dynamic scenarios. As dynamic graph generation progresses amid evolving latent non-stationary environments, it is imperative to explore their effects on out-of-distribution (OOD) generalization. This paper proposes a novel Evolving Graph Learning framework for OOD generalization (EvoOOD) by environment-aware invariant pattern recognition. Specifically, we first design an environment sequential variational auto-encoder to model environment evolution and infer the underlying environment distribution. Then, we introduce a mechanism for environment-aware invariant pattern recognition, tailored to address environmental diversification through inferred distributions. Finally, we conduct fine-grained causal interventions on individual nodes using a mixture of instantiated environment samples. This approach helps to distinguish spatio-temporal invariant patterns for OOD prediction, especially in non-stationary environments. Experimental results demonstrate the superiority of EvoGOOD on both real-world and synthetic dynamic datasets under distribution shifts. To the best of our knowledge, it is the first attempt to study the dynamic graph OOD generalization problem from the environment evolution perspective.