Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Fully Hyperbolic Neural Networks

Weize Chen, Xu Han, Yankai Lin, Hexu Zhao, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou


Abstract
Hyperbolic neural networks have shown great potential for modeling complex data. However, existing hyperbolic networks are not completely hyperbolic, as they encode features in the hyperbolic space yet formalize most of their operations in the tangent space (a Euclidean subspace) at the origin of the hyperbolic model. This hybrid method greatly limits the modeling ability of networks. In this paper, we propose a fully hyperbolic framework to build hyperbolic networks based on the Lorentz model by adapting the Lorentz transformations (including boost and rotation) to formalize essential operations of neural networks. Moreover, we also prove that linear transformation in tangent spaces used by existing hyperbolic networks is a relaxation of the Lorentz rotation and does not include the boost, implicitly limiting the capabilities of existing hyperbolic networks. The experimental results on four NLP tasks show that our method has better performance for building both shallow and deep networks. Our code will be released to facilitate follow-up research.
Anthology ID:
2022.acl-long.389
Volume:
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
May
Year:
2022
Address:
Dublin, Ireland
Editors:
Smaranda Muresan, Preslav Nakov, Aline Villavicencio
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
5672–5686
Language:
URL:
https://aclanthology.org/2022.acl-long.389
DOI:
10.18653/v1/2022.acl-long.389
Bibkey:
Cite (ACL):
Weize Chen, Xu Han, Yankai Lin, Hexu Zhao, Zhiyuan Liu, Peng Li, Maosong Sun, and Jie Zhou. 2022. Fully Hyperbolic Neural Networks. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5672–5686, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
Fully Hyperbolic Neural Networks (Chen et al., ACL 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.acl-long.389.pdf
Code
 chenweize1998/fully-hyperbolic-nn
Data
FB15k-237