Machine-Learned Sampling of Conditioned Path Measures
Qijia Jiang, Reuben Cohn-Gordon · 2025
Read Machine-Learned Sampling of Conditioned Path Measures by Qijia Jiang, Reuben Cohn-Gordon (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-2506-01904.txt.
Description
Jiang and Cohn-Gordon propose machine-learned ways to sample whole paths from a posterior distribution. Many problems ask not for one path but for the spread of paths that fit a condition. Their methods borrow from controlled dynamics that morph one path law into another. They also use geometry on probability space, with a Wasserstein metric. That geometry lets a density curve bend toward the target likelihood.
The algorithms rest on theory rather than data. Neural networks can be plugged in to learn the target path families. No sample data are needed to drive the learning. The result is a sampler for conditioned paths under a general prior process.
Catalog record
| CREATOR | Qijia Jiang, Reuben Cohn-Gordon |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 87 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-01904