Higher-Order Explanations of Graph Neural Networks via Relevant Walks

By Thomas Schnake, Oliver Eberle, Jonas Lederer, Shinichi Nakajima, Kristof T. Schütt, Klaus‐Robert Müller , Grgoire Montavon. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020

310 citations recorded by the source corpus.

Graph Neural Networks (GNNs) are a popular approach for predicting graph structured data. As GNNs tightly entangle the input graph into the neural network structure, common explainable AI approaches are not applicable. To a large extent, GNNs have remained black-boxes for the user so far. In this paper, we show that GNNs can in fact be naturally explained using higher-order expansions, i.e., by identifying groups of edges that jointly contribute to the prediction. Practically, we find that such explanations can be extracted using a nested attribution scheme, where existing techniques such as layer-wise relevance propagation (LRP) can be applied at each step. The output is a collection of walks into the input graph that are relevant for the prediction. Our novel explanation method, which we denote by GNN-LRP, is applicable to a broad range of graph neural networks and lets us extract practically relevant insights on sentiment analysis of text data, structure-property relationships in quantum chemistry, and image classification.

Subfields: Artificial Intelligence, Materials Chemistry

View Paper: https://www.semanticscholar.org/paper/9e707dd89bba25a3dd22c96f43bd72b9b3ab94bb

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Materials Chemistry

Citations in the corpus

Explainability in Graph Neural Networks: A Taxonomic Survey (2020, methodology): “library, we include the implementations of several existing methods, such as GNNExplainer [46], PGExplainer [47], DeepLIFT [64], GNN-LRP [60], Grad-CAM [55], SubgraphX”
Interpreting Graph Neural Networks for NLP With Differentiable Edge Masking (2020, background): “Similarly, Schwarzenberg et al. (2019); Baldassarre & Azizpour (2019); Schnake et al. (2020) apply Layerwise Relevance Propagation (Bach et al., 2015) to the GNN setting.”
Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications (2021, methodology): “At the same time, these methods were able to handle different types of data, including images [13], speech [20], text [10], [38], and structured data, such as molecules [162], [165] or genetic sequences [191].”
Graph Neural Networks in Recommender Systems: A Survey (2020, background): “Themethods can be divided into two categories: the instance-levelmethods provide example-specific explanations by identifying important input features for its prediction [3, 123, 194]; the model-level methods provide high-level interpretations and a generic understanding of how deep graph models wor”