Thomas Kipf

Kipf co-developed graph convolutional networks and variational graph autoencoders, influential methods for learning from graph-structured data. At Google DeepMind he studies object-centric representations, relational reasoning, and models that capture the compositional structure of physical and digital worlds.