Entropy Aware Message Passing in Graph Neural Networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nazari, Philipp, Lemke, Oliver, Guidobene, Davide, Gesp, Artiom
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909131466604544
author Nazari, Philipp
Lemke, Oliver
Guidobene, Davide
Gesp, Artiom
author_facet Nazari, Philipp
Lemke, Oliver
Guidobene, Davide
Gesp, Artiom
contents Deep Graph Neural Networks struggle with oversmoothing. This paper introduces a novel, physics-inspired GNN model designed to mitigate this issue. Our approach integrates with existing GNN architectures, introducing an entropy-aware message passing term. This term performs gradient ascent on the entropy during node aggregation, thereby preserving a certain degree of entropy in the embeddings. We conduct a comparative analysis of our model against state-of-the-art GNNs across various common datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2403_04636
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Entropy Aware Message Passing in Graph Neural Networks
Nazari, Philipp
Lemke, Oliver
Guidobene, Davide
Gesp, Artiom
Machine Learning
I.2.6; I.5.1
Deep Graph Neural Networks struggle with oversmoothing. This paper introduces a novel, physics-inspired GNN model designed to mitigate this issue. Our approach integrates with existing GNN architectures, introducing an entropy-aware message passing term. This term performs gradient ascent on the entropy during node aggregation, thereby preserving a certain degree of entropy in the embeddings. We conduct a comparative analysis of our model against state-of-the-art GNNs across various common datasets.
title Entropy Aware Message Passing in Graph Neural Networks
topic Machine Learning
I.2.6; I.5.1
url https://arxiv.org/abs/2403.04636