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.
Описание
[1806.08804] Hierarchical Graph Representation Learning with Differentiable Pooling
%0 Generic
%1 ying2018hierarchical
%A Ying, Rex
%A You, Jiaxuan
%A Morris, Christopher
%A Ren, Xiang
%A Hamilton, William L.
%A Leskovec, Jure
%D 2018
%K hierarchical plk
%T Hierarchical Graph Representation Learning with Differentiable Pooling
%U http://arxiv.org/abs/1806.08804
%X 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.
@misc{ying2018hierarchical,
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.},
added-at = {2021-09-13T15:01:19.000+0200},
author = {Ying, Rex and You, Jiaxuan and Morris, Christopher and Ren, Xiang and Hamilton, William L. and Leskovec, Jure},
biburl = {https://www.bibsonomy.org/bibtex/2bb0b743eeb85232c8a6019c890448e8a/parismic},
description = {[1806.08804] Hierarchical Graph Representation Learning with Differentiable Pooling},
interhash = {b3874b1dd2e71e675538db8bd3e5d7eb},
intrahash = {bb0b743eeb85232c8a6019c890448e8a},
keywords = {hierarchical plk},
note = {cite arxiv:1806.08804},
timestamp = {2021-09-13T15:01:19.000+0200},
title = {Hierarchical Graph Representation Learning with Differentiable Pooling},
url = {http://arxiv.org/abs/1806.08804},
year = 2018
}