Contrastive Graph Neural Network Explanation
arXiv:2010.13663
Abstract
Graph Neural Networks achieve remarkable results on problems with structured data but come as black-box predictors. Transferring existing explanation techniques, such as occlusion, fails as even removing a single node or edge can lead to drastic changes in the graph. The resulting graphs can differ from all training examples, causing model confusion and wrong explanations. Thus, we argue that explicability must use graphs compliant with the distribution underlying the training data. We coin this property Distribution Compliant Explanation (DCE) and present a novel Contrastive GNN Explanation (CoGE) technique following this paradigm. An experimental study supports the efficacy of CoGE.
arXiv commentICML 2020 Workshop on Graph Representation Learning and Beyond (GRL+)
References in corpus (9)
- Adam: A Method for Stochastic Optimization
- Inductive Representation Learning on Large Graphs
- Neural Message Passing for Quantum Chemistry
- Graph Neural Networks: A Review of Methods and Applications
- Adversarial Attacks on Neural Networks for Graph Data
- Representation Learning on Graphs with Jumping Knowledge Networks
- Adversarial Attacks on Graph Neural Networks via Meta Learning
- Interpolating between Optimal Transport and MMD using Sinkhorn Divergences
- Explainability Techniques for Graph Convolutional Networks
Cited by in corpus (5)
- Differentially Private Graph Classification with GNNs
- A Meta-Learning Approach for Training Explainable Graph Neural Networks
- GRETEL: A unified framework for Graph Counterfactual Explanation Evaluation
- An Empirical Study: Extensive Deep Temporal Point Process
- Self-learn to Explain Siamese Networks Robustly