Hierarchical Graph Representation Learning with Differentiable Pooling

TL;DR

This paper introduces DiffPool, a differentiable pooling method for GNNs that learns hierarchical graph representations, significantly improving graph classification accuracy and setting new benchmarks.

Contribution

The paper presents DiffPool, a novel differentiable pooling technique enabling hierarchical graph representations within GNNs for the first time.

Findings

Achieved 5-10% accuracy improvements on graph classification benchmarks.

Set new state-of-the-art results on four out of five datasets.

Demonstrated effective integration of DiffPool with various GNN architectures.

Abstract

Recently, graph neural networks (GNNs) have revolutionized the field of graph representation learning through effectively learned node embeddings, and achieved state-of-the-art results in tasks such as node classification and link prediction. However, current GNN methods are inherently flat and do not learn hierarchical representations of graphs---a limitation that is especially problematic for the task of graph classification, where the goal is to predict the label associated with an entire graph. Here we propose DiffPool, a differentiable graph pooling module that can generate hierarchical representations of graphs and can be combined with various graph neural network architectures in an end-to-end fashion. DiffPool learns a differentiable soft cluster assignment for nodes at each layer of a deep GNN, mapping nodes to a set of clusters, which then form the coarsened input for the next GNN layer. Our experimental results show that combining existing GNN methods with DiffPool yields an average improvement of 5-10% accuracy on graph classification benchmarks, compared to all existing pooling approaches, achieving a new state-of-the-art on four out of five benchmark data sets.