The aim of this paper is to give a systematic mathematical interpretation of the diffusion problem on which Graph Neural Networks (GNNs) models are based. The starting point of our approach is a dissipative functional leading to dynamical equations which allows us to study the symmetries of the model. We discuss the conserved charges and provide a charge-preserving numerical method for solving the dynamical equations. In any dynamical system and also in GRAph Neural Diffusion (GRAND), knowing the charge values and their conservation along the evolution flow could provide a way to understand how GNNs and other networks work with their learning capabilities.
A charge-preserving method for solving graph neural diffusion networks
Lidia Aceto
;Pietro Grassi
2024-01-01
Abstract
The aim of this paper is to give a systematic mathematical interpretation of the diffusion problem on which Graph Neural Networks (GNNs) models are based. The starting point of our approach is a dissipative functional leading to dynamical equations which allows us to study the symmetries of the model. We discuss the conserved charges and provide a charge-preserving numerical method for solving the dynamical equations. In any dynamical system and also in GRAph Neural Diffusion (GRAND), knowing the charge values and their conservation along the evolution flow could provide a way to understand how GNNs and other networks work with their learning capabilities.File | Dimensione | Formato | |
---|---|---|---|
2312.10279.pdf
file ad accesso aperto
Descrizione: pre-print
Tipologia:
Documento in Pre-print
Licenza:
Dominio pubblico
Dimensione
260.63 kB
Formato
Adobe PDF
|
260.63 kB | Adobe PDF | Visualizza/Apri |
2025_CNSNS.pdf
file ad accesso aperto
Descrizione: article
Tipologia:
Versione Editoriale (PDF)
Licenza:
Non specificato
Dimensione
845.23 kB
Formato
Adobe PDF
|
845.23 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.