Guardado en:
Detalles Bibliográficos
Autores principales: Roth, Andreas, Liebig, Thomas
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:https://arxiv.org/abs/2308.16800
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916398188462080
author Roth, Andreas
Liebig, Thomas
author_facet Roth, Andreas
Liebig, Thomas
contents Our study reveals new theoretical insights into over-smoothing and feature over-correlation in graph neural networks. Specifically, we demonstrate that with increased depth, node representations become dominated by a low-dimensional subspace that depends on the aggregation function but not on the feature transformations. For all aggregation functions, the rank of the node representations collapses, resulting in over-smoothing for particular aggregation functions. Our study emphasizes the importance for future research to focus on rank collapse rather than over-smoothing. Guided by our theory, we propose a sum of Kronecker products as a beneficial property that provably prevents over-smoothing, over-correlation, and rank collapse. We empirically demonstrate the shortcomings of existing models in fitting target functions of node classification tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2308_16800
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Rank Collapse Causes Over-Smoothing and Over-Correlation in Graph Neural Networks
Roth, Andreas
Liebig, Thomas
Machine Learning
Artificial Intelligence
Our study reveals new theoretical insights into over-smoothing and feature over-correlation in graph neural networks. Specifically, we demonstrate that with increased depth, node representations become dominated by a low-dimensional subspace that depends on the aggregation function but not on the feature transformations. For all aggregation functions, the rank of the node representations collapses, resulting in over-smoothing for particular aggregation functions. Our study emphasizes the importance for future research to focus on rank collapse rather than over-smoothing. Guided by our theory, we propose a sum of Kronecker products as a beneficial property that provably prevents over-smoothing, over-correlation, and rank collapse. We empirically demonstrate the shortcomings of existing models in fitting target functions of node classification tasks.
title Rank Collapse Causes Over-Smoothing and Over-Correlation in Graph Neural Networks
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2308.16800