Saved in:
Bibliographic Details
Main Authors: Tai, Wenxin, Liu, Yaqian, Zhong, Ting, Zhou, Fan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.07527
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911735622926336
author Tai, Wenxin
Liu, Yaqian
Zhong, Ting
Zhou, Fan
author_facet Tai, Wenxin
Liu, Yaqian
Zhong, Ting
Zhou, Fan
contents Recent work has observed that explanations produced by Self-Interpretable Graph Neural Networks (SI-GNNs) can be self-inconsistent: when the model is reapplied to its own explanatory graph subset, it may produce a different explanation. However, why self-inconsistency arises remains poorly understood. In this work, we first identify re-explanation-induced context perturbation as the direct cause of score variation. We then introduce a latent signal assignment hypothesis to explain why only some edges are sensitive to this perturbation, and analyze how conciseness regularization affects latent signal assignment. Given that self-inconsistent edges do not provide stable evidence for the model's prediction, we propose Self-Denoising (SD), a model-agnostic and training-free post-processing strategy that calibrates explanations with only one additional forward pass. Experiments across representative SI-GNN frameworks, backbone architectures, and benchmark datasets support our hypothesis and show that SD consistently improves explanation quality while adding only about 4--6\% computational overhead in practice.
format Preprint
id arxiv_https___arxiv_org_abs_2605_07527
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Why Self-Inconsistency Arises in GNN Explanations and How to Exploit It
Tai, Wenxin
Liu, Yaqian
Zhong, Ting
Zhou, Fan
Machine Learning
Artificial Intelligence
Recent work has observed that explanations produced by Self-Interpretable Graph Neural Networks (SI-GNNs) can be self-inconsistent: when the model is reapplied to its own explanatory graph subset, it may produce a different explanation. However, why self-inconsistency arises remains poorly understood. In this work, we first identify re-explanation-induced context perturbation as the direct cause of score variation. We then introduce a latent signal assignment hypothesis to explain why only some edges are sensitive to this perturbation, and analyze how conciseness regularization affects latent signal assignment. Given that self-inconsistent edges do not provide stable evidence for the model's prediction, we propose Self-Denoising (SD), a model-agnostic and training-free post-processing strategy that calibrates explanations with only one additional forward pass. Experiments across representative SI-GNN frameworks, backbone architectures, and benchmark datasets support our hypothesis and show that SD consistently improves explanation quality while adding only about 4--6\% computational overhead in practice.
title Why Self-Inconsistency Arises in GNN Explanations and How to Exploit It
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2605.07527