Plug-In Classification of Drift Functions in Diffusion Processes Using Neural Networks
Yuzhen Zhao, Jiarong Fan, Yating Liu · 2026
Read Plug-In Classification of Drift Functions in Diffusion Processes Using Neural Networks by Yuzhen Zhao, Jiarong Fan, Yating Liu (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-2602-02791.txt.
Description
Yuzhen Zhao and 2 co-authors classify which diffusion process generated an observed trajectory by first learning each class's drift, in research posted to arXiv in 2026. Each class is defined by a distinct drift function, the steady pull inside the random motion, and trajectories are sampled at discrete times.
The authors first derive the multidimensional Bayes rule for the problem. Their plug-in classifier fits each drift with neural networks, then applies that rule. Convergence rates for the excess misclassification risk separate the parts played by drift estimation, time discretization, and dimension. Exploiting diffusion structure helps: the drift is learned from every observed increment, sharpening guarantees over classifiers fed raw paths. Tests beat earlier methods in one dimension, stay effective in higher dimensions when drifts have compositional structure, and beat end-to-end trajectory classifiers.
Catalog record
| CREATOR | Yuzhen Zhao, Jiarong Fan, Yating Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-02791