Witness Autoencoder: Shaping the Latent Space with Witness Complexes

Abstract

We present a Witness Autoencoder (W-AE) – an autoencoder that captures geodesic distances of the data in the latent space. Our algorithm uses witness complexes to compute geodesic distance approximations on a mini-batch level, and leverages topological information from the entire dataset while performing batch-wise approximations. This way, our method allows to capture the global structure of the data even with a small batch size, which is beneficial for large-scale real-world data. We show that our method captures the structure of the manifold more accurately than the recently introduced topological autoencoder (TopoAE).

Cite

Text

Schönenberger et al. "Witness Autoencoder: Shaping the Latent Space with Witness Complexes ." NeurIPS 2020 Workshops: TDA_and_Beyond, 2020.

Markdown

[Schönenberger et al. "Witness Autoencoder: Shaping the Latent Space with Witness Complexes ." NeurIPS 2020 Workshops: TDA_and_Beyond, 2020.](https://mlanthology.org/neuripsw/2020/schonenberger2020neuripsw-witness/)

BibTeX

@inproceedings{schonenberger2020neuripsw-witness,
  title     = {{Witness Autoencoder: Shaping the Latent Space with Witness Complexes }},
  author    = {Schönenberger, Simon Till and Varava, Anastasiia and Polianskii, Vladislav and Chung, Jen Jen and Kragic, Danica and Siegwart, Roland},
  booktitle = {NeurIPS 2020 Workshops: TDA_and_Beyond},
  year      = {2020},
  url       = {https://mlanthology.org/neuripsw/2020/schonenberger2020neuripsw-witness/}
}