Hyperbolic Attention Networks
Abstract
We introduce hyperbolic attention networks to endow neural networks with enough capacity to match the complexity of data with hierarchical and power-law structure. A few recent approaches have successfully demonstrated the benefits of imposing hyperbolic geometry on the parameters of shallow networks. We extend this line of work by imposing hyperbolic geometry on the activations of neural networks. This allows us to exploit hyperbolic geometry to reason about embeddings produced by deep networks. We achieve this by re-expressing the ubiquitous mechanism of soft attention in terms of operations defined for hyperboloid and Klein models. Our method shows improvements in terms of generalization on neural machine translation, learning on graphs and visual question answering tasks while keeping the neural representations compact.
Cite
@article{arxiv.1805.09786,
title = {Hyperbolic Attention Networks},
author = {Caglar Gulcehre and Misha Denil and Mateusz Malinowski and Ali Razavi and Razvan Pascanu and Karl Moritz Hermann and Peter Battaglia and Victor Bapst and David Raposo and Adam Santoro and Nando de Freitas},
journal= {arXiv preprint arXiv:1805.09786},
year = {2018}
}