MPXGAT: An Attention based Deep Learning Model for Multiplex Graphs Embedding

Marco Bongiovanni, Luca Gallo, Roberto Grasso, Alfredo Pulvirenti

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: Multiplex Graphs, GAT, Link prediction, Multiplex embedding

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: We propose MPXGAT, an attention-based deep learning model for embedding multiplex graphs that exploits intra-layer and inter-layer connections to enable link prediction tasks within and across different layers.

Abstract:

Graph representation learning is a research area that has attracted a lot of attention in recent years.
However, most of the existing studies focus on the embedding of single-layer graphs, which cannot describe systems where multiple types of relationships exist.
Here we propose MPXGAT, an attention-based deep learning model for embedding multiplex graphs.
Our methodology, which is based on GATs, embeds the nodes of a multiplex network by exploiting both their intra-layer and inter-layer connections, enabling link prediction tasks within and across different layers.
A thorough experimental analysis on three benchmark datasets, shows that MPXGAT outperforms state-of-the-art competing algorithms.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 5291