Hyperspherical Variational Auto-Encoders

Tim Davidson, Luca Falorsi, Nicola De Cao, Thomas Kipf, Jakub M. Tomczak
Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, PMLR R16:855-864, 2018.

Abstract

The Variational Auto-Encoder (VAE) is one of the most used unsupervised machine learn- ing models. But although the default choice of a Gaussian distribution for both the prior and posterior represents a mathematically con- venient distribution often leading to competi- tive results, we show that this parameterization fails to model data with a latent hyperspheri- cal structure. To address this issue we propose using a von Mises-Fisher (vMF) distribution in- stead, leading to a hyperspherical latent space. Through a series of experiments we show how such a hyperspherical VAE, or S-VAE, is more suitable for capturing data with a hyperspheri- cal latent structure, while outperforming a nor- mal, N-VAE, in low dimensions on other data types.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR16-davidson18a, title = {Hyperspherical Variational Auto-Encoders}, author = {Davidson, Tim and Falorsi, Luca and De Cao, Nicola and Kipf, Thomas and Tomczak, Jakub M.}, booktitle = {Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence}, pages = {855--864}, year = {2018}, editor = {Globerson, Amir and Silva, Ricardo}, volume = {R16}, series = {Proceedings of Machine Learning Research}, month = {06--10 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r16/main/assets/davidson18a/davidson18a.pdf}, url = {https://proceedings.mlr.press/r16/davidson18a.html}, abstract = {The Variational Auto-Encoder (VAE) is one of the most used unsupervised machine learn- ing models. But although the default choice of a Gaussian distribution for both the prior and posterior represents a mathematically con- venient distribution often leading to competi- tive results, we show that this parameterization fails to model data with a latent hyperspheri- cal structure. To address this issue we propose using a von Mises-Fisher (vMF) distribution in- stead, leading to a hyperspherical latent space. Through a series of experiments we show how such a hyperspherical VAE, or S-VAE, is more suitable for capturing data with a hyperspheri- cal latent structure, while outperforming a nor- mal, N-VAE, in low dimensions on other data types.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Hyperspherical Variational Auto-Encoders %A Tim Davidson %A Luca Falorsi %A Nicola De Cao %A Thomas Kipf %A Jakub M. Tomczak %B Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2018 %E Amir Globerson %E Ricardo Silva %F pmlr-vR16-davidson18a %I PMLR %P 855--864 %U https://proceedings.mlr.press/r16/davidson18a.html %V R16 %X The Variational Auto-Encoder (VAE) is one of the most used unsupervised machine learn- ing models. But although the default choice of a Gaussian distribution for both the prior and posterior represents a mathematically con- venient distribution often leading to competi- tive results, we show that this parameterization fails to model data with a latent hyperspheri- cal structure. To address this issue we propose using a von Mises-Fisher (vMF) distribution in- stead, leading to a hyperspherical latent space. Through a series of experiments we show how such a hyperspherical VAE, or S-VAE, is more suitable for capturing data with a hyperspheri- cal latent structure, while outperforming a nor- mal, N-VAE, in low dimensions on other data types. %Z Reissued by PMLR on 04 October 2026.
APA
Davidson, T., Falorsi, L., De Cao, N., Kipf, T. & Tomczak, J.M.. (2018). Hyperspherical Variational Auto-Encoders. Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R16:855-864 Available from https://proceedings.mlr.press/r16/davidson18a.html. Reissued by PMLR on 04 October 2026.

Related Material