Overcoming Fake Solutions in Semi-Dual Neural Optimal Transport: A Smoothing Approach for Learning the Optimal Transport Plan
Jaemoo Choi, Jaewoong Choi, Dohyun Kwon · 2025
Read Overcoming Fake Solutions in Semi-Dual Neural Optimal Transport: A Smoothing Approach for Learning the Optimal Transport Plan by Jaemoo Choi, Jaewoong Choi, Dohyun Kwon (2025) 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-2502-04583.txt.
Description
Jaemoo Choi, Jaewoong Choi, and Dohyun Kwon fix fake solutions in neural optimal transport in a 2025 paper. Moving one distribution onto another at least cost is a core task in imaging and learning. A popular semi-dual neural method often converges to maps that look trained but fail to actually carry the source onto the target.
The authors first state a clean condition under which the method's answer is the true transport map. When the condition fails, their new model learns both the map and the full transport plan. Under sharp assumptions they prove the fake solutions disappear. Tests confirm recovery where rivals fail, including one-to-many colorization.
Catalog record
| CREATOR | Jaemoo Choi, Jaewoong Choi, Dohyun Kwon |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 76 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-04583