How Faithful are Self-Explainable GNNs?
arXiv:2308.15096
Abstract
Self-explainable deep neural networks are a recent class of models that can output ante-hoc local explanations that are faithful to the model's reasoning, and as such represent a step forward toward filling the gap between expressiveness and interpretability. Self-explainable graph neural networks (GNNs) aim at achieving the same in the context of graph data. This begs the question: do these models fulfill their implicit guarantees in terms of faithfulness? In this extended abstract, we analyze the faithfulness of several self-explainable GNNs using different measures of faithfulness, identify several limitations -- both in the models themselves and in the evaluation metrics -- and outline possible ways forward.
References in corpus (15)
- Statistical mechanics of complex networks
- GNNExplainer: Generating Explanations for Graph Neural Networks
- This Looks Like That: Deep Learning for Interpretable Image Recognition
- TUDataset: A collection of benchmark datasets for learning with graphs
- Concept Whitening for Interpretable Image Recognition
- Parameterized Explainer for Graph Neural Network
- Concept Bottleneck Models
- The Bayesian Case Model: A Generative Approach for Case-Based Reasoning and Prototype Classification
- Interpretable and Generalizable Graph Learning via Stochastic Attention Mechanism
- GraphFramEx: Towards Systematic Evaluation of Explainability Methods for Graph Neural Networks
- A Survey on Explainability of Graph Neural Networks
- Concept Embedding Analysis: A Review
- Global Explainability of GNNs via Logic Combination of Learned Concepts
- Towards Prototype-Based Self-Explainable Graph Neural Network
- GlanceNets: Interpretabile, Leak-proof Concept-based Models