Metraipolis › archive › dc_eac8b661d78e
Beyond Uniform Local Isometry and Topology: FactoMap for Disentangled Representations — arXiv:2608.24762
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.24762
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T06:36:25.142Z
- SHA-256 of the archived text
f84a4531b1b5e9ff15a74abc75630bd3b6d189541c189cc8015b317e0f3d264a- Stable URL of this page
https://metraipolis.com/documento/dc_eac8b661d78e
How to cite
Beyond Uniform Local Isometry and Topology: FactoMap for Disentangled Representations — arXiv:2608.24762 — arXiv (en), https://arxiv.org/abs/2608.24762 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_eac8b661d78e (Metraipolis archive dc_eac8b661d78e, sha256 f84a4531b1b5e9ff15a74abc75630bd3b6d189541c189cc8015b317e0f3d264a)
BibTeX
@misc{dceac8b661d78e,
title = {Beyond Uniform Local Isometry and Topology: FactoMap for Disentangled Representations — arXiv:2608.24762},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.24762},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_eac8b661d78e; sha256 f84a4531b1b5e9ff15a74abc75630bd3b6d189541c189cc8015b317e0f3d264a},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.24762. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Many disentanglement methods represent generative factors using Euclidean product coordinates, although the underlying factor spaces may wrap, collapse, or have position-dependent geometry. We introduce factor-space structure, combining factor domains, generator-induced identifications, and position-dependent scales to distinguish topologically equivalent spaces with different factor geometries. We show that statistically independent factors need not be geometrically separable: hue and scale produce effects that grow at different rates, yielding anisotropy that no fixed rescaling removes. We propose the Factor-Space Topographic Map (FactoMap), which learns interpretable prototypes indexed by a factor-space lattice. Topographic learning transfers the lattice's periodicity, collapses, and non-uniform extent to the representation. Experiments show that matching this structure preserves factor continuity and enables disentanglement of the underlying factors. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-26 in cs.LG). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.24762 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-18T06:28:52.280Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-26?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).