Riemannian Neural Optimal Transport
Alessandro Micheli, Yueqi Cao, Anthea Monod, Samir Bhatt · 2026
Read Riemannian Neural Optimal Transport by Alessandro Micheli, Yueqi Cao, Anthea Monod, Samir Bhatt (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-03566.txt.
Description
Optimal transport moves mass from one distribution to another at minimal cost, and neural versions learn that map from data. Alessandro Micheli and 3 co-authors extend the idea to curved spaces in a 2026 paper. Existing neural transport methods assume flat Euclidean geometry, while many real datasets live on Riemannian manifolds.
First the authors prove a barrier: any method that discretizes transport maps on manifolds needs exponentially many parameters as dimension grows. Their answer, Riemannian Neural OT, builds the map with a neural network that carries the geometry by construction. Under mild assumptions it comes close to true transport maps with sub-exponential complexity. Experiments on synthetic and real data show better scaling than discretization baselines.
Catalog record
| CREATOR | Alessandro Micheli, Yueqi Cao, Anthea Monod, Samir Bhatt |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 188 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-03566