GNNExplainer: Generating Explanations for Graph Neural Networks (2019-03-10T00:00:00.000000Z)

TL;DR

GnExplainer is proposed, the first general, model-agnostic approach for providing interpretable explanations for predictions of any GNN-based model on any graph-based machine learning task.

Abstract

Graph Neural Networks (GNNs) are a powerful tool for machine learning on graphs. GNNs combine node feature information with the graph structure by recursively passing neural messages along edges of the input graph. However, incorporating both graph structure and feature information leads to complex models and explaining predictions made by GNNs remains unsolved. Here we propose GnnExplainer, the first general, model-agnostic approach for providing interpretable explanations for predictions of any GNN-based model on any graph-based machine learning task. Given an instance, GnnExplainer identifies a compact subgraph structure and a small subset of node features that have a crucial role in GNN's prediction. Further, GnnExplainer can generate consistent and concise explanations for an entire class of instances. We formulate GnnExplainer as an optimization task that maximizes the mutual information between a GNN's prediction and distribution of possible subgraph structures. Experiments on synthetic and real-world graphs show that our approach can identify important graph structures as well as node features, and outperforms alternative baseline approaches by up to 43.0% in explanation accuracy. GnnExplainer provides a variety of benefits, from the ability to visualize semantically relevant structures to interpretability, to giving insights into errors of faulty GNNs.

Authors

M. Zitnik

10 papers

J. Leskovec

38 papers

Rex Ying

7 papers

TL;DR

Abstract

Authors

References62 items

Position-aware Graph Neural Networks

ExplaiNE: An Approach for Explaining Network Embedding-based Link Predictions

Graph Neural Networks: A Review of Methods and Applications

Interpretable Graph Convolutional Neural Networks for Inference on Noisy Knowledge Graphs

Representer Point Selection for Explaining Deep Neural Networks

Sanity Checks for Saliency Maps

How Powerful are Graph Neural Networks?

Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)

Adaptive Sampling Towards Fast Graph Representation Learning

Hierarchical Graph Representation Learning with Differentiable Pooling

Representation Learning on Graphs with Jumping Knowledge Networks

Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation

Graph Convolutional Neural Networks for Web-Scale Recommender Systems

Relational inductive biases, deep learning, and graph networks

Community Interaction and Conflict on the Web

Link Prediction Based on Graph Neural Networks

Learning to Explain: An Information-Theoretic Perspective on Model Interpretation

Neural Relational Inference for Interacting Systems

A Survey of Methods for Explaining Black Box Models

Modeling polypharmacy side effects with graph convolutional networks

FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling

All Models are Wrong but many are Useful: Variable Importance for Black-Box, Proprietary, or Misspecified Prediction Models, using Model Class Reliance

Graph Attention Networks

Stochastic Training of Graph Convolutional Networks with Variance Reduction

Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties.

Representation Learning on Graphs: Methods and Applications

Large-Scale Analysis of Disease Pathways in the Human Interactome

Predicting Organic Reaction Outcomes with Weisfeiler-Lehman Network

Interpretable & Explorable Approximations of Black Box Models

Explanation in Artificial Intelligence: Insights from the Social Sciences

Inductive Representation Learning on Large Graphs

Supervised Community Detection with Line Graph Neural Networks

A Unified Approach to Interpreting Model Predictions

Learning Important Features Through Propagating Activation Differences

Neural Message Passing for Quantum Chemistry

Modeling Relational Data with Graph Convolutional Networks

Understanding Black-box Predictions via Influence Functions

Axiomatic Attribution for Deep Networks

Towards A Rigorous Science of Interpretable Machine Learning

Visualizing Deep Neural Network Decisions: Prediction Difference Analysis

DeepRED - Rule Extraction from Deep Neural Networks

Semi-Supervised Classification with Graph Convolutional Networks

Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Discriminative Embeddings of Latent Variable Models for Structured Data

“Why Should I Trust You?”: Explaining the Predictions of Any Classifier

Gated Graph Sequence Neural Networks

Convolutional Networks on Graphs for Learning Molecular Fingerprints

Deep Graph Kernels

Auto-Encoding Variational Bayes

Visualizing and Understanding Convolutional Networks

Reverse Engineering the Neural Networks for Rule Extraction in Classification Problems

Friendship and mobility: user movement in location-based social networks

Supervised random walks: predicting and recommending links in social networks

Discovering additive structure in black box functions

Structure-activity relationship of mutagenic aromatic and heteroaromatic nitro compounds. Correlation with molecular orbital energies and hydrophobicity.

and s

Rex Ying

Examples are not enough, learn to criticize! Criticism for Interpretability

ANN-DT : An Algorithm for Extraction of Decision Trees from Artificial Neural Networks

I and J

Visualizing Higher-Layer Features of a Deep Network

The Graph Neural Network Model

Field of Study

Journal Information

Name

Page

Volume

Venue Information

Name

Type

URL

Alternate Names