Work

MagNet: A Neural Network for Directed Graphs

Year: 2021

Type: article

Abstract: , a GNN for directed graphs based on a complex Hermitian matrix known as the magnetic Laplacian. This matrix encodes undirected geometric structure in the magnitude of its entries and directional information in their phase. A "charge" parameter attunes spectral information to variation among directed cycles. We apply our network to a variety of directed graph node classification and link prediction tasks showing that MagNet performs well on all tasks and that its performance exceeds all other methods on a majority of such tasks. The underlying principles of MagNet are such that it can be adapted to other GNN architectures.

Source: PubMed

DOI: 10.48550/arxiv.2102.11391

Authors: Zhang, Xitong, He, Yixuan, Brugnone, Nathan, Perlmutter, Michael, Hirn, Matthew

Institutions: Michigan State University, University of Oxford, Science Oxford, University of California, Los Angeles

Language: English

Topic: Advanced Graph Neural Networks

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: National Institute of General Medical Sciences

Award: Finding emergent structure in multi-sample biological data with the dual geometry of cells and features