Visualizing Graph Neural Networks with CorGIE: Corresponding a Graph to Its Embedding

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liu, Zipeng, Wang, Yang, Bernard, Jürgen, Munzner, Tamara
Natura: Preprint
Pubblicazione: 2021
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908435062194176
author Liu, Zipeng
Wang, Yang
Bernard, Jürgen
Munzner, Tamara
author_facet Liu, Zipeng
Wang, Yang
Bernard, Jürgen
Munzner, Tamara
contents Graph neural networks (GNNs) are a class of powerful machine learning tools that model node relations for making predictions of nodes or links. GNN developers rely on quantitative metrics of the predictions to evaluate a GNN, but similar to many other neural networks, it is difficult for them to understand if the GNN truly learns characteristics of a graph as expected. We propose an approach to corresponding an input graph to its node embedding (aka latent space), a common component of GNNs that is later used for prediction. We abstract the data and tasks, and develop an interactive multi-view interface called CorGIE to instantiate the abstraction. As the key function in CorGIE, we propose the K-hop graph layout to show topological neighbors in hops and their clustering structure. To evaluate the functionality and usability of CorGIE, we present how to use CorGIE in two usage scenarios, and conduct a case study with five GNN experts.
format Preprint
id arxiv_https___arxiv_org_abs_2106_12839
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Visualizing Graph Neural Networks with CorGIE: Corresponding a Graph to Its Embedding
Liu, Zipeng
Wang, Yang
Bernard, Jürgen
Munzner, Tamara
Machine Learning
Human-Computer Interaction
Graph neural networks (GNNs) are a class of powerful machine learning tools that model node relations for making predictions of nodes or links. GNN developers rely on quantitative metrics of the predictions to evaluate a GNN, but similar to many other neural networks, it is difficult for them to understand if the GNN truly learns characteristics of a graph as expected. We propose an approach to corresponding an input graph to its node embedding (aka latent space), a common component of GNNs that is later used for prediction. We abstract the data and tasks, and develop an interactive multi-view interface called CorGIE to instantiate the abstraction. As the key function in CorGIE, we propose the K-hop graph layout to show topological neighbors in hops and their clustering structure. To evaluate the functionality and usability of CorGIE, we present how to use CorGIE in two usage scenarios, and conduct a case study with five GNN experts.
title Visualizing Graph Neural Networks with CorGIE: Corresponding a Graph to Its Embedding
topic Machine Learning
Human-Computer Interaction
url https://arxiv.org/abs/2106.12839