Geometric Deep Graph Learning: Exploring Opportunities in Different Geometric Spaces

Explore the cutting-edge field of geometric deep graph learning in this illuminating one-hour conference talk by Philip S. Yu at #ICBS2024. Delve into the fundamental question of which representation spaces are most suitable for complex graph structures beyond traditional Euclidean space. Examine the fascinating properties of alternative geometric spaces like hyperbolic and hyperspherical, and discover how they can enhance various graph learning tasks including classification, clustering, contrastive learning, graph structure learning, and continual graph learning. Gain insights into the potential of these approaches to revolutionize deep graph learning for applications in social networks, transportation systems, financial networks, and biochemical structures.