Graph Neural Networks
Petar discusses the flexibility of message functions in graph neural networks, highlighting the importance of weight sharing between isomorphic graph structures. This approach challenges the traditional view of group theory in geometric deep learning, introducing the concept of groupoids for enhanced model scalability and function building.In this clip
From this podcast
Machine Learning Street Talk (MLST)
#85 Dr. Petar Veličković (Deepmind) - Categories, Graphs, Reasoning [NEURIPS22 UNPLUGGED]
Related Questions