The article introduces BLIS-Net, a novel Graph Neural Network (GNN) designed for signal classification. Unlike traditional GNNs, BLIS-Net can capture both local and global signal structures, as well as low and high-frequency information. This makes it superior for tasks such as classifying and analyzing signals on graphs, including real-world data sets based on traffic flow and fMRI data. The authors argue that BLIS-Net’s design and capabilities make it an effective tool for tasks that have received less attention in traditional GNN research.
Publication date: 26 Oct 2023
Project Page: https://arxiv.org/abs/2310.17579v1
Paper: https://arxiv.org/pdf/2310.17579
