Poincar\'e Wasserstein Autoencoder

Ivan Ovinnikov

Poincar\'e Wasserstein Autoencoder

Machine Learning 2020-03-18 v2 Machine Learning

Authors: Ivan Ovinnikov

Abstract

This work presents a reformulation of the recently proposed Wasserstein autoencoder framework on a non-Euclidean manifold, the Poincar\'e ball model of the hyperbolic space. By assuming the latent space to be hyperbolic, we can use its intrinsic hierarchy to impose structure on the learned latent space representations. We demonstrate the model in the visual domain to analyze some of its properties and show competitive results on a graph link prediction task.

Keywords

hyperbolic embeddings manifold learning

Cite

@article{arxiv.1901.01427,
  title  = {Poincar\'e Wasserstein Autoencoder},
  author = {Ivan Ovinnikov},
  journal= {arXiv preprint arXiv:1901.01427},
  year   = {2020}
}

Related papers

View all related →

Machine Learning · Statistics

Continuous Hierarchical Representations with Poincar\'e Variational Auto-Encoders

Emile Mathieu, Charline Le Lan, Chris J. Maddison, Ryota Tomioka +1

2019-11-27

Artificial Intelligence · Computer Science

Poincar\'e Embeddings for Learning Hierarchical Representations