LGAINov 5, 2024

Dissecting the Failure of Invariant Learning on Graphs

arXiv:2411.02847v36 citationsh-index: 14Has CodeNIPS
Originality Incremental advance
AI Analysis

This addresses a crucial issue for graph machine learning by enhancing OOD generalization, though it is incremental as it builds on prior invariant learning methods.

The paper tackles the problem of node-level Out-Of-Distribution (OOD) generalization on graphs by analyzing the failure of existing invariant learning methods and proposing Cross-environment Intra-class Alignment (CIA) and its variant CIA-LRA, which improve performance on graph OOD benchmarks.

Enhancing node-level Out-Of-Distribution (OOD) generalization on graphs remains a crucial area of research. In this paper, we develop a Structural Causal Model (SCM) to theoretically dissect the performance of two prominent invariant learning methods -- Invariant Risk Minimization (IRM) and Variance-Risk Extrapolation (VREx) -- in node-level OOD settings. Our analysis reveals a critical limitation: due to the lack of class-conditional invariance constraints, these methods may struggle to accurately identify the structure of the predictive invariant ego-graph and consequently rely on spurious features. To address this, we propose Cross-environment Intra-class Alignment (CIA), which explicitly eliminates spurious features by aligning cross-environment representations conditioned on the same class, bypassing the need for explicit knowledge of the causal pattern structure. To adapt CIA to node-level OOD scenarios where environment labels are hard to obtain, we further propose CIA-LRA (Localized Reweighting Alignment) that leverages the distribution of neighboring labels to selectively align node representations, effectively distinguishing and preserving invariant features while removing spurious ones, all without relying on environment labels. We theoretically prove CIA-LRA's effectiveness by deriving an OOD generalization error bound based on PAC-Bayesian analysis. Experiments on graph OOD benchmarks validate the superiority of CIA and CIA-LRA, marking a significant advancement in node-level OOD generalization. The codes are available at https://github.com/NOVAglow646/NeurIPS24-Invariant-Learning-on-Graphs.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes