Zorro: Valid, Sparse, and Stable Explanations in Graph Neural Networks

With the ever-increasing popularity and applications of graph neural networks, several proposals have been made to interpret and understand the decisions of a GNN model. Explanations for a GNN model differ in principle from other input settings. It is important to attribute the decision to input features and other related instances connected by the graph structure.We find that the previous explanation generation approaches that maximize the mutual information between the label distribution produced by the GNN model and the explanation to be restrictive. Specifically, existing approaches do not enforce explanations to be predictive, sparse, or robust to input perturbations. In this paper, we lay down some of the fundamental principles that an explanation method for GNNs should follow and introduce a metric fidelity as a measure of the explanation’s effectiveness. We propose a novel approach Zorro based on the principles from rate-distortion theory that uses a simple combinatorial procedure to optimize for fidelity. Extensive experiments on real and synthetic datasets reveal that Zorro produces sparser, stable, andmore faithful explanations than existing GNN explanation approaches. ACM Reference Format: Thorben Funke, Megha Khosla, and Avishek Anand. 2018. Zorro: Valid, Sparse, and Stable Explanations in Graph Neural Networks. InWoodstock ’18: ACM Symposium on Neural Gaze Detection, June 03–05, 2018, Woodstock, NY . ACM,NewYork, NY, USA, 12 pages. https://doi.org/10.1145/1122445.1122456

[1]  Avanti Shrikumar,et al.  Learning Important Features Through Propagating Activation Differences , 2017, ICML.

[2]  Jonas Mueller,et al.  What made you do this? Understanding black-box decisions with sufficient input subsets , 2018, AISTATS.

[3]  Le Song,et al.  Learning to Explain: An Information-Theoretic Perspective on Model Interpretation , 2018, ICML.

[4]  Dumitru Erhan,et al.  A Benchmark for Interpretability Methods in Deep Neural Networks , 2018, NeurIPS.

[5]  Max Welling,et al.  Semi-Supervised Classification with Graph Convolutional Networks , 2016, ICLR.

[6]  Lior Wolf,et al.  A Formal Approach to Explainability , 2019, AIES.

[7]  Ruslan Salakhutdinov,et al.  Revisiting Semi-Supervised Learning with Graph Embeddings , 2016, ICML.

[8]  My T. Thai,et al.  PGM-Explainer: Probabilistic Graphical Model Explanations for Graph Neural Networks , 2020, NeurIPS.

[9]  Megha Khosla,et al.  Finding Interpretable Concept Spaces in Node Embeddings using Knowledge Bases , 2019, PKDD/ECML Workshops.

[10]  Steven Skiena,et al.  DeepWalk: online learning of social representations , 2014, KDD.

[11]  Dumitru Erhan,et al.  The (Un)reliability of saliency methods , 2017, Explainable AI.

[12]  Shuiwang Ji,et al.  XGNN: Towards Model-Level Explanations of Graph Neural Networks , 2020, KDD.

[13]  Tijl De Bie,et al.  ExplaiNE: An Approach for Explaining Network Embedding-based Link Predictions , 2019, ArXiv.

[14]  Jan Eric Lenssen,et al.  Fast Graph Representation Learning with PyTorch Geometric , 2019, ArXiv.

[15]  Thomas Lukasiewicz,et al.  The Struggles of Feature-Based Explanations: Shapley Values vs. Minimal Sufficient Subsets , 2020, ArXiv.

[16]  Yoav Goldberg,et al.  Aligning Faithful Interpretations with their Social Attribution , 2020, ArXiv.

[17]  Heiko Hoffmann,et al.  Explainability Methods for Graph Convolutional Neural Networks , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[18]  C. Sims.  Rate–distortion theory and human perception , 2016, Cognition.

[19]  Jure Leskovec,et al.  GNN Explainer: A Tool for Post-hoc Explanation of Graph Neural Networks , 2019, ArXiv.

[20]  Yoshua Bengio,et al.  Show, Attend and Tell: Neural Image Caption Generation with Visual Attention , 2015, ICML.

[21]  Graham Neubig,et al.  Learning to Deceive with Attention-Based Explanations , 2020, ACL.

[22]  Carlos Guestrin,et al.  Anchors: High-Precision Model-Agnostic Explanations , 2018, AAAI.

[23]  M. de Rijke,et al.  Do Transformer Attention Heads Provide Transparency in Abstractive Summarization? , 2019, ArXiv.

[24]  Jure Leskovec,et al.  How Powerful are Graph Neural Networks? , 2018, ICLR.

[25]  Carlos Guestrin,et al.  "Why Should I Trust You?": Explaining the Predictions of Any Classifier , 2016, ArXiv.

[26]  Stephan Günnemann,et al.  Predict then Propagate: Graph Neural Networks meet Personalized PageRank , 2018, ICLR.

[27]  Lucy J. Colwell,et al.  Evaluating Attribution for Graph Neural Networks , 2020, NeurIPS.

[28]  Alexander Binder,et al.  Layer-Wise Relevance Propagation for Neural Networks with Local Renormalization Layers , 2016, ICANN.

[29]  Megha Khosla,et al.  A Comparative Study for Unsupervised Network Representation Learning , 2019 .

[30]  Ankur Taly,et al.  Axiomatic Attribution for Deep Networks , 2017, ICML.