Evaluating Neighbor Explainability for Graph Neural Networks

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Llorente, Oscar, Fawzy, Rana, Keown, Jared, Horemuz, Michal, Vaderna, Péter, Laki, Sándor, Kotroczó, Roland, Csoma, Rita, Szalai-Gindl, János Márk
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914735204597760
author Llorente, Oscar
Fawzy, Rana
Keown, Jared
Horemuz, Michal
Vaderna, Péter
Laki, Sándor
Kotroczó, Roland
Csoma, Rita
Szalai-Gindl, János Márk
author_facet Llorente, Oscar
Fawzy, Rana
Keown, Jared
Horemuz, Michal
Vaderna, Péter
Laki, Sándor
Kotroczó, Roland
Csoma, Rita
Szalai-Gindl, János Márk
contents Explainability in Graph Neural Networks (GNNs) is a new field growing in the last few years. In this publication we address the problem of determining how important is each neighbor for the GNN when classifying a node and how to measure the performance for this specific task. To do this, various known explainability methods are reformulated to get the neighbor importance and four new metrics are presented. Our results show that there is almost no difference between the explanations provided by gradient-based techniques in the GNN domain. In addition, many explainability techniques failed to identify important neighbors when GNNs without self-loops are used.
format Preprint
id arxiv_https___arxiv_org_abs_2311_08118
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Evaluating Neighbor Explainability for Graph Neural Networks
Llorente, Oscar
Fawzy, Rana
Keown, Jared
Horemuz, Michal
Vaderna, Péter
Laki, Sándor
Kotroczó, Roland
Csoma, Rita
Szalai-Gindl, János Márk
Machine Learning
Artificial Intelligence
Explainability in Graph Neural Networks (GNNs) is a new field growing in the last few years. In this publication we address the problem of determining how important is each neighbor for the GNN when classifying a node and how to measure the performance for this specific task. To do this, various known explainability methods are reformulated to get the neighbor importance and four new metrics are presented. Our results show that there is almost no difference between the explanations provided by gradient-based techniques in the GNN domain. In addition, many explainability techniques failed to identify important neighbors when GNNs without self-loops are used.
title Evaluating Neighbor Explainability for Graph Neural Networks
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2311.08118