Rethinking Graph Out-Of-Distribution Generalization: A Learnable Random Walk Perspective

2025-05-09

AI Summary

Graph neural networks (GNNs) suffer significant performance degradation under distributional shift, as existing out-of-distribution (OOD) generalization methods rely on fixed graph topologies or spectral structures to enforce invariance—assumptions misaligned with real-world distributional changes. This paper proposes a learnable random walk (LRW) perspective, modeling cross-distribution invariant knowledge as data-driven walk dynamics, thereby abandoning rigid structural assumptions. Our approach features two core innovations: (1) a jointly parameterized LRW-sampler and path encoder that learns node transition matrices end-to-end; and (2) a mutual information maximization loss based on kernel density estimation (KDE), explicitly enforcing discriminability and robustness of walk paths under OOD conditions. Evaluated across multiple graph OOD benchmarks, our method achieves an average accuracy improvement of 3.87% over state-of-the-art approaches.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Out-Of-Distribution (OOD) generalization has gained increasing attentions for machine learning on graphs, as graph neural networks (GNNs) often exhibit performance degradation under distribution shifts. Existing graph OOD methods tend to follow the basic ideas of invariant risk minimization and structural causal models, interpreting the invariant knowledge across datasets under various distribution shifts as graph topology or graph spectrum. However, these interpretations may be inconsistent with real-world scenarios, as neither invariant topology nor spectrum is assured. In this paper, we advocate the learnable random walk (LRW) perspective as the instantiation of invariant knowledge, and propose LRW-OOD to realize graph OOD generalization learning. Instead of employing fixed probability transition matrix (i.e., degree-normalized adjacency matrix), we parameterize the transition matrix with an LRW-sampler and a path encoder. Furthermore, we propose the kernel density estimation (KDE)-based mutual information (MI) loss to generate random walk sequences that adhere to OOD principles. Extensive experiment demonstrates that our model can effectively enhance graph OOD generalization under various types of distribution shifts and yield a significant accuracy improvement of 3.87% over state-of-the-art graph OOD generalization baselines.

Problem

• Addressing performance degradation of GNNs under graph distribution shifts
• Proposing learnable random walks for invariant knowledge in OOD generalization
• Enhancing OOD generalization via KDE-based mutual information loss

Innovation

• Learnable random walk for graph OOD generalization
• Parameterized transition matrix with LRW-sampler
• KDE-based mutual information loss for OOD principles