Improving global awareness of linkset predictions using Cross-Attentive Modulation tokens

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Marcoccia, Félix, Adjih, Cédric, Mühlethaler, Paul
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912035909926912
author Marcoccia, Félix
Adjih, Cédric
Mühlethaler, Paul
author_facet Marcoccia, Félix
Adjih, Cédric
Mühlethaler, Paul
contents This work introduces Cross-Attentive Modulation (CAM) tokens, which are tokens whose initial value is learned, gather information through cross-attention, and modulate the nodes and edges accordingly. These tokens are meant to improve the global awareness of link predictions models which, based on graph neural networks, can struggle to capture graph-level features. This lack of ability to feature high level representations is particularly limiting when predicting multiple or entire sets of links. We implement CAM tokens in a simple attention-based link prediction model and in a graph transformer, which we also use in a denoising diffusion framework. A brief introduction to our toy datasets will then be followed by benchmarks which prove that CAM token improve the performance of the model they supplement and outperform a baseline with diverse statistical graph attributes.
format Preprint
id arxiv_https___arxiv_org_abs_2405_19375
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Improving global awareness of linkset predictions using Cross-Attentive Modulation tokens
Marcoccia, Félix
Adjih, Cédric
Mühlethaler, Paul
Social and Information Networks
Machine Learning
I.2.6
This work introduces Cross-Attentive Modulation (CAM) tokens, which are tokens whose initial value is learned, gather information through cross-attention, and modulate the nodes and edges accordingly. These tokens are meant to improve the global awareness of link predictions models which, based on graph neural networks, can struggle to capture graph-level features. This lack of ability to feature high level representations is particularly limiting when predicting multiple or entire sets of links. We implement CAM tokens in a simple attention-based link prediction model and in a graph transformer, which we also use in a denoising diffusion framework. A brief introduction to our toy datasets will then be followed by benchmarks which prove that CAM token improve the performance of the model they supplement and outperform a baseline with diverse statistical graph attributes.
title Improving global awareness of linkset predictions using Cross-Attentive Modulation tokens
topic Social and Information Networks
Machine Learning
I.2.6
url https://arxiv.org/abs/2405.19375