Submission Deadline: 31 December 2024 View: 400 Submit to Special Issue
Graph Neural Networks (GNNs) have emerged as a cutting-edge framework for directly analyzing graph-structured data using deep learning techniques. Their remarkable performance has inspired extensive exploration and research interest among scholars. GNNs excel in transforming graph data into standardized representations, enabling them to effectively address tasks such as node classification, edge information propagation, and graph clustering. What distinguishes GNNs from traditional deep learning methods is their distinctive capacity to capture intrinsic patterns and deeper semantic features within graphs. This unique capability markedly boosts their accuracy and robustness across a wide spectrum of graph-related challenges, establishing them as a highly promising approach in contemporary machine learning.
The proposed special issue aims to explore the diverse methods and applications of GNNs across multiple domains, including materials, transportation, meteorology, and beyond. By addressing challenges and harnessing the potential of GNNs, this special issue endeavors to uncover innovative approaches to enhance GNNs, providing a platform to integrate domain-specific knowledge and fully exploit their potential in real-world scenarios across diverse domains.
The potential topics include but are not limited to:
Heterogeneous Graphs and Their Applications
Dynamic Graphs and Their Applications
Hypergraphs and Their Applications
Spatio-Temporal Graphs and Their Applications
Multi-Task GNNs and Their Applications
Explainability on GNNs and Their Applications
Generative and Pre-trained Models for GNNs and Their Applications
Large-scale Graphs and Their Applications
Graph Foundation Models and Their Applications