Desiderata for Representation Learning from Identifiability, Disentanglement, and Group-Structuredness
Abstract
Machine learning subfields define useful representations differently: disentanglement strives for semantic meaning and symmetries, identifiability for recovering the ground-truth factors of the (unobservable) data generating process, group-structured representations for equivariance. We demonstrate that despite their merits, each approach has shortcomings. Surprisingly, joining forces helps overcome the limitations: we use insights from latent space statistics, geometry, and topology in our examples to elucidate how combining the desiderata of identifiability, disentanglement, and group structure yields more useful representations.
Cite
Text
Keurti et al. "Desiderata for Representation Learning from Identifiability, Disentanglement, and Group-Structuredness." ICML 2023 Workshops: TAGML, 2023.Markdown
[Keurti et al. "Desiderata for Representation Learning from Identifiability, Disentanglement, and Group-Structuredness." ICML 2023 Workshops: TAGML, 2023.](https://mlanthology.org/icmlw/2023/keurti2023icmlw-desiderata/)BibTeX
@inproceedings{keurti2023icmlw-desiderata,
title = {{Desiderata for Representation Learning from Identifiability, Disentanglement, and Group-Structuredness}},
author = {Keurti, Hamza and Reizinger, Patrik and Schölkopf, Bernhard and Brendel, Wieland},
booktitle = {ICML 2023 Workshops: TAGML},
year = {2023},
url = {https://mlanthology.org/icmlw/2023/keurti2023icmlw-desiderata/}
}