Improving Subgraph Recognition with Variational Graph Information Bottleneck

12/18/2021
by   Junchi Yu, et al.
6

Subgraph recognition aims at discovering a compressed substructure of a graph that is most informative to the graph property. It can be formulated by optimizing Graph Information Bottleneck (GIB) with a mutual information estimator. However, GIB suffers from training instability since the mutual information of graph data is intrinsically difficult to estimate. This paper introduces a noise injection method to compress the information in the subgraphs, which leads to a novel Variational Graph Information Bottleneck (VGIB) framework. VGIB allows a tractable variational approximation to its objective under mild assumptions. Therefore, VGIB enjoys more stable and efficient training process - we find that VGIB converges 10 times faster than GIB with improved performances in practice. Extensive experiments on graph interpretation, explainability of Graph Neural Networks, and graph classification show that VGIB finds better subgraphs than existing methods.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
10/12/2020

Graph Information Bottleneck for Subgraph Recognition

Given the input graph and its label/property, several key problems of gr...
research
03/20/2021

Recognizing Predictive Substructures with Subgraph Information Bottleneck

The emergence of Graph Convolutional Network (GCN) has greatly boosted t...
research
01/20/2021

SUGAR: Subgraph Neural Network with Reinforcement Pooling and Self-Supervised Mutual Information Mechanism

Graph representation learning has attracted increasing research attentio...
research
12/30/2021

Deconfounded Training for Graph Neural Networks

Learning powerful representations is one central theme of graph neural n...
research
07/14/2022

Subgraph Frequency Distribution Estimation using Graph Neural Networks

Small subgraphs (graphlets) are important features to describe fundament...
research
04/29/2021

Learning Robust Variational Information Bottleneck with Reference

We propose a new approach to train a variational information bottleneck ...
research
12/02/2022

AGO: Boosting Mobile AI Inference Performance by Removing Constraints on Graph Optimization

Traditional deep learning compilers rely on heuristics for subgraph gene...

Please sign up or login with your details

Forgot password? Click here to reset