A Meta-Learning Approach for Training Explainable Graph Neural Networks
arXiv:2109.09426 · doi:10.1109/TNNLS.2022.3171398
Abstract
In this paper, we investigate the degree of explainability of graph neural networks (GNNs). Existing explainers work by finding global/local subgraphs to explain a prediction, but they are applied after a GNN has already been trained. Here, we propose a meta-learning framework for improving the level of explainability of a GNN directly at training time, by steering the optimization procedure towards what we call `interpretable minima'. Our framework (called MATE, MetA-Train to Explain) jointly trains a model to solve the original task, e.g., node classification, and to provide easily processable outputs for downstream algorithms that explain the model's decisions in a human-friendly way. In particular, we meta-train the model's parameters to quickly minimize the error of an instance-level GNNExplainer trained on-the-fly on randomly sampled nodes. The final internal representation relies upon a set of features that can be `better' understood by an explanation algorithm, e.g., another instance of GNNExplainer. Our model-agnostic approach can improve the explanations produced for different GNN architectures and use any instance-based explainer to drive this process. Experiments on synthetic and real-world datasets for node and graph classification show that we can produce models that are consistently easier to explain by different algorithms. Furthermore, this increase in explainability comes at no cost for the accuracy of the model.
References in corpus (19)
- Adam: A Method for Stochastic Optimization
- Statistical mechanics of complex networks
- A Comprehensive Survey on Graph Neural Networks
- Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
- Neural Message Passing for Quantum Chemistry
- GNNExplainer: Generating Explanations for Graph Neural Networks
- Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
- A Gentle Introduction to Deep Learning for Graphs
- XGNN: Towards Model-Level Explanations of Graph Neural Networks
- Parameterized Explainer for Graph Neural Network
- Interpretable and Efficient Heterogeneous Graph Convolutional Network
- Adaptive Propagation Graph Convolutional Network
- Explainability in Graph Neural Networks: A Taxonomic Survey
- Generalized Inner Loop Meta-Learning
- GraphLIME: Local Interpretable Model Explanations for Graph Neural Networks
- CF-GNNExplainer: Counterfactual Explanations for Graph Neural Networks
- Interpreting Graph Neural Networks for NLP With Differentiable Edge Masking
- Medical Entity Disambiguation Using Graph Neural Networks
- Contrastive Graph Neural Network Explanation