On Evaluation Metrics for Graph Generative Models

In image generation, generative models can be evaluated naturally by visually inspecting model outputs. However, this is not always the case for graph generative models (GGMs), making their evaluation challenging. Currently, the standard process for evaluating GGMs suffers from three critical limita...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Thompson, Rylee, Knyazev, Boris, Ghalebi, Elahe, Kim, Jungtaek, Taylor, Graham W
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Thompson, Rylee
Knyazev, Boris
Ghalebi, Elahe
Kim, Jungtaek
Taylor, Graham W
description In image generation, generative models can be evaluated naturally by visually inspecting model outputs. However, this is not always the case for graph generative models (GGMs), making their evaluation challenging. Currently, the standard process for evaluating GGMs suffers from three critical limitations: i) it does not produce a single score which makes model selection challenging, ii) in many cases it fails to consider underlying edge and node features, and iii) it is prohibitively slow to perform. In this work, we mitigate these issues by searching for scalar, domain-agnostic, and scalable metrics for evaluating and ranking GGMs. To this end, we study existing GGM metrics and neural-network-based metrics emerging from generative models of images that use embeddings extracted from a task-specific network. Motivated by the power of certain Graph Neural Networks (GNNs) to extract meaningful graph representations without any training, we introduce several metrics based on the features extracted by an untrained random GNN. We design experiments to thoroughly test metrics on their ability to measure the diversity and fidelity of generated graphs, as well as their sample and computational efficiency. Depending on the quantity of samples, we recommend one of two random-GNN-based metrics that we show to be more expressive than pre-existing metrics. While we focus on applying these metrics to GGM evaluation, in practice this enables the ability to easily compute the dissimilarity between any two sets of graphs regardless of domain. Our code is released at: https://github.com/uoguelph-mlrg/GGM-metrics.
doi_str_mv 10.48550/arxiv.2201.09871
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2201_09871</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2201_09871</sourcerecordid><originalsourceid>FETCH-LOGICAL-a671-99899fa916f5a37c8f6b8d49100fe32f48955c4c7bb060943718e23cb418eb9a3</originalsourceid><addsrcrecordid>eNotzr0KwjAYheEsDqJegJO5gdakSZp8o4hWQXFxL19igoXaSqpF797f6R0OHB5Cppyl0ijF5hgfVZ9mGeMpA6P5kPBDQ1c91ne8VW1D9_4WK9fR0EZaRLyeaeEbH99j7-m-Pfm6G5NBwLrzk39H5LheHZebZHcotsvFLsFc8wTAAAQEngeFQjsTcmtOEjhjwYssSANKOem0tSxnIIXmxmfCWfmuBRQjMvvdfs3lNVYXjM_yYy-_dvECFO89Ww</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>On Evaluation Metrics for Graph Generative Models</title><source>arXiv.org</source><creator>Thompson, Rylee ; Knyazev, Boris ; Ghalebi, Elahe ; Kim, Jungtaek ; Taylor, Graham W</creator><creatorcontrib>Thompson, Rylee ; Knyazev, Boris ; Ghalebi, Elahe ; Kim, Jungtaek ; Taylor, Graham W</creatorcontrib><description>In image generation, generative models can be evaluated naturally by visually inspecting model outputs. However, this is not always the case for graph generative models (GGMs), making their evaluation challenging. Currently, the standard process for evaluating GGMs suffers from three critical limitations: i) it does not produce a single score which makes model selection challenging, ii) in many cases it fails to consider underlying edge and node features, and iii) it is prohibitively slow to perform. In this work, we mitigate these issues by searching for scalar, domain-agnostic, and scalable metrics for evaluating and ranking GGMs. To this end, we study existing GGM metrics and neural-network-based metrics emerging from generative models of images that use embeddings extracted from a task-specific network. Motivated by the power of certain Graph Neural Networks (GNNs) to extract meaningful graph representations without any training, we introduce several metrics based on the features extracted by an untrained random GNN. We design experiments to thoroughly test metrics on their ability to measure the diversity and fidelity of generated graphs, as well as their sample and computational efficiency. Depending on the quantity of samples, we recommend one of two random-GNN-based metrics that we show to be more expressive than pre-existing metrics. While we focus on applying these metrics to GGM evaluation, in practice this enables the ability to easily compute the dissimilarity between any two sets of graphs regardless of domain. Our code is released at: https://github.com/uoguelph-mlrg/GGM-metrics.</description><identifier>DOI: 10.48550/arxiv.2201.09871</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Learning</subject><creationdate>2022-01</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2201.09871$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2201.09871$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Thompson, Rylee</creatorcontrib><creatorcontrib>Knyazev, Boris</creatorcontrib><creatorcontrib>Ghalebi, Elahe</creatorcontrib><creatorcontrib>Kim, Jungtaek</creatorcontrib><creatorcontrib>Taylor, Graham W</creatorcontrib><title>On Evaluation Metrics for Graph Generative Models</title><description>In image generation, generative models can be evaluated naturally by visually inspecting model outputs. However, this is not always the case for graph generative models (GGMs), making their evaluation challenging. Currently, the standard process for evaluating GGMs suffers from three critical limitations: i) it does not produce a single score which makes model selection challenging, ii) in many cases it fails to consider underlying edge and node features, and iii) it is prohibitively slow to perform. In this work, we mitigate these issues by searching for scalar, domain-agnostic, and scalable metrics for evaluating and ranking GGMs. To this end, we study existing GGM metrics and neural-network-based metrics emerging from generative models of images that use embeddings extracted from a task-specific network. Motivated by the power of certain Graph Neural Networks (GNNs) to extract meaningful graph representations without any training, we introduce several metrics based on the features extracted by an untrained random GNN. We design experiments to thoroughly test metrics on their ability to measure the diversity and fidelity of generated graphs, as well as their sample and computational efficiency. Depending on the quantity of samples, we recommend one of two random-GNN-based metrics that we show to be more expressive than pre-existing metrics. While we focus on applying these metrics to GGM evaluation, in practice this enables the ability to easily compute the dissimilarity between any two sets of graphs regardless of domain. Our code is released at: https://github.com/uoguelph-mlrg/GGM-metrics.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotzr0KwjAYheEsDqJegJO5gdakSZp8o4hWQXFxL19igoXaSqpF797f6R0OHB5Cppyl0ijF5hgfVZ9mGeMpA6P5kPBDQ1c91ne8VW1D9_4WK9fR0EZaRLyeaeEbH99j7-m-Pfm6G5NBwLrzk39H5LheHZebZHcotsvFLsFc8wTAAAQEngeFQjsTcmtOEjhjwYssSANKOem0tSxnIIXmxmfCWfmuBRQjMvvdfs3lNVYXjM_yYy-_dvECFO89Ww</recordid><startdate>20220124</startdate><enddate>20220124</enddate><creator>Thompson, Rylee</creator><creator>Knyazev, Boris</creator><creator>Ghalebi, Elahe</creator><creator>Kim, Jungtaek</creator><creator>Taylor, Graham W</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20220124</creationdate><title>On Evaluation Metrics for Graph Generative Models</title><author>Thompson, Rylee ; Knyazev, Boris ; Ghalebi, Elahe ; Kim, Jungtaek ; Taylor, Graham W</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a671-99899fa916f5a37c8f6b8d49100fe32f48955c4c7bb060943718e23cb418eb9a3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Thompson, Rylee</creatorcontrib><creatorcontrib>Knyazev, Boris</creatorcontrib><creatorcontrib>Ghalebi, Elahe</creatorcontrib><creatorcontrib>Kim, Jungtaek</creatorcontrib><creatorcontrib>Taylor, Graham W</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Thompson, Rylee</au><au>Knyazev, Boris</au><au>Ghalebi, Elahe</au><au>Kim, Jungtaek</au><au>Taylor, Graham W</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>On Evaluation Metrics for Graph Generative Models</atitle><date>2022-01-24</date><risdate>2022</risdate><abstract>In image generation, generative models can be evaluated naturally by visually inspecting model outputs. However, this is not always the case for graph generative models (GGMs), making their evaluation challenging. Currently, the standard process for evaluating GGMs suffers from three critical limitations: i) it does not produce a single score which makes model selection challenging, ii) in many cases it fails to consider underlying edge and node features, and iii) it is prohibitively slow to perform. In this work, we mitigate these issues by searching for scalar, domain-agnostic, and scalable metrics for evaluating and ranking GGMs. To this end, we study existing GGM metrics and neural-network-based metrics emerging from generative models of images that use embeddings extracted from a task-specific network. Motivated by the power of certain Graph Neural Networks (GNNs) to extract meaningful graph representations without any training, we introduce several metrics based on the features extracted by an untrained random GNN. We design experiments to thoroughly test metrics on their ability to measure the diversity and fidelity of generated graphs, as well as their sample and computational efficiency. Depending on the quantity of samples, we recommend one of two random-GNN-based metrics that we show to be more expressive than pre-existing metrics. While we focus on applying these metrics to GGM evaluation, in practice this enables the ability to easily compute the dissimilarity between any two sets of graphs regardless of domain. Our code is released at: https://github.com/uoguelph-mlrg/GGM-metrics.</abstract><doi>10.48550/arxiv.2201.09871</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2201.09871
ispartof
issn
language eng
recordid cdi_arxiv_primary_2201_09871
source arXiv.org
subjects Computer Science - Artificial Intelligence
Computer Science - Learning
title On Evaluation Metrics for Graph Generative Models
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T20%3A53%3A11IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=On%20Evaluation%20Metrics%20for%20Graph%20Generative%20Models&rft.au=Thompson,%20Rylee&rft.date=2022-01-24&rft_id=info:doi/10.48550/arxiv.2201.09871&rft_dat=%3Carxiv_GOX%3E2201_09871%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true