NICE: Scale-Stable Perturbations for Graph Neural Network Explanations via Noise Corruption

2026-08-16

AI Summary

This study addresses scale drift and distribution shift caused by element-wise mask perturbations in GNN explainers. We propose a scale-stable noise erosion mechanism to replace traditional masking, thereby preserving message norm stability, alongside a stochastic recovery boundary and gradient integration attribution method. This framework eliminates scale drift while significantly enhancing the reliability and faithfulness of model explanations. Experimental results demonstrate that our approach effectively mitigates the instability inherent in existing explainers, establishing a new paradigm for trustworthy attribution in graph neural networks. Extensive evaluations across multiple benchmarks confirm superior performance compared to state-of-the-art methods, validating both the theoretical soundness and practical efficacy of the proposed framework for robust GNN interpretability.

Abstract

Post-hoc Graph Neural Network (GNN) explainers commonly follow a Perturb-Query paradigm, inferring the importance of graph elements based on queried predictions to perturbed inputs. However, such perturbations often introduce substantial distribution shift, undermining the reliability of the queried predictions used to derive explanations. While existing efforts mainly improve perturbed graphs or stabilize model predictions on them, we revisit the perturbation mechanism itself. We show that the widely used Element-wise Masking(EM) suppresses edge-induced messages toward zero, causing deterministic scale contraction that accumulates across message-passing layers, a phenomenon we term Scale Drift. Consequently, prediction changes under EM may conflate information corruption with deviations in propagation scale. As a scale-stable alternative to EM, we introduce Noise Corruption (NC), which perturbs each message through matched-norm random-direction corruption while preserving the expected squared message norm. Building on NC, we propose NICE, a Noise Corruption-based explanation framework, which learns a Stochastic Restoration Boundary (SRB) under NC-induced uncertainty, balancing target-prediction restoration against compactness. Furthermore, Boundary-Integrated Gradient (BIG) converts this boundary into edge attributions by accumulating each edge's contribution to reducing restoration risk along the restoration path. Experiments across multiple benchmarks demonstrate stronger explanation performance and model faithfulness while confirming that NC substantially reduces the Scale Drift induced by masking.

Problem

• Graph Neural Network Explanations
• Scale Drift
• Element-wise Masking
• Distribution Shift
• Perturb-Query Paradigm

Innovation

• Scale-Stable Perturbations
• Noise Corruption
• Scale Drift
• Stochastic Restoration Boundary
• Boundary-Integrated Gradient