`

Timezone: »

 
Poster
Hyperbolic Graph Neural Networks
Qi Liu · Maximilian Nickel · Douwe Kiela

Wed Dec 11 05:00 PM -- 07:00 PM (PST) @ East Exhibition Hall B + C #65
in Algorithms -- Relational Learning »

Learning from graph-structured data is an important task in machine learning and artificial intelligence, for which Graph Neural Networks (GNNs) have shown great promise. Motivated by recent advances in geometric representation learning, we propose a novel GNN architecture for learning representations on Riemannian manifolds with differentiable exponential and logarithmic maps. We develop a scalable algorithm for modeling the structural properties of graphs, comparing Euclidean and hyperbolic geometry. In our experiments, we show that hyperbolic GNNs can lead to substantial improvements on various benchmark datasets.

Keywords: Representation Learning · Relational Learning · Algorithms
[ Paper]  [ Poster] 

Author Information

Qi Liu (University of Oxford)
Maximilian Nickel (Facebook AI Research)
Douwe Kiela (Facebook AI Research)

More from the Same Authors