Metraipolis › archive › dc_096ff514fb61
Shared Geometry As A Rosetta Stone: Cross-Modal Alignment Without Paired Data — arXiv:2610.09411
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.09411
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T11:10:58.283Z
- SHA-256 of the archived text
69a169da7c43380e2a83fb3b176de4121f08ae25f7396549e420617adfff64ee- Stable URL of this page
https://metraipolis.com/documento/dc_096ff514fb61
How to cite
Shared Geometry As A Rosetta Stone: Cross-Modal Alignment Without Paired Data — arXiv:2610.09411 — arXiv (en), https://arxiv.org/abs/2610.09411 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_096ff514fb61 (Metraipolis archive dc_096ff514fb61, sha256 69a169da7c43380e2a83fb3b176de4121f08ae25f7396549e420617adfff64ee)
BibTeX
@misc{dc096ff514fb61,
title = {Shared Geometry As A Rosetta Stone: Cross-Modal Alignment Without Paired Data — arXiv:2610.09411},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.09411},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_096ff514fb61; sha256 69a169da7c43380e2a83fb3b176de4121f08ae25f7396549e420617adfff64ee},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.09411. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multimodal representations enable zero-shot classification and retrieval, but aligning independently trained models usually requires large amounts of paired data. Yet, the Platonic Representation Hypothesis suggests that models trained on different modalities may converge spontaneously toward a shared representation geometry. But then, do we even need paired examples for cross-modal alignment? Remarkably, we show that paired examples are unnecessary for coarse cross-modal alignment. Our simple Wasserstein Procrustes method with a coarse geometric initialization aligns two disjoint embedding sets by estimating a single orthogonal map without seeing any pairs. Across datasets, modalities, and unimodal models, we show that we can consistently align independently trained representations without pairs, and standard geometric alignment metrics accurately predict when this is possible. Nevertheless, we can naturally benefit from paired examples. In the very few-pair regime, our method substantially outperforms existing ones, while staying competitive with pair-based methods with more added examples. Finally, we demonstrate that the resulting alignments can enable text-to-image generation without paired examples. These results show that independently trained models often share enough geometry to establish cross-modal correspondence with little or no paired data. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2610.09411 — 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 2026-10-08T10:14:06.083Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).