Regular Fourier Features for Nonstationary Gaussian Processes
Arsalan Jawaid, Abdullah Karatas, Jörg Seewig · 2026
Read Regular Fourier Features for Nonstationary Gaussian Processes by Arsalan Jawaid, Abdullah Karatas, Jörg Seewig (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-2602-23006.txt.
Description
Arsalan Jawaid and two co-authors propose regular Fourier features for shifting Gaussian processes. Simulating a Gaussian process means drawing from a high-dimensional Gaussian spread, which scales cubically with sample spots. Spectral methods ease this by treating the spectral density as a chance spread for Monte Carlo use, but this reading fails for shifting processes, where densities are mostly not chance measures.
The new method slices the spectral form directly for harmony-type processes with one-dimensional inputs, keeping link structure among spectral weights without chance guesses. With finite spectral support it gives a fast low-rank estimate, valid by design and steady under mild smoothness rules. The setup also extends to kernel learning from data, matching strong baselines on real and synthetic tests.
Catalog record
| CREATOR | Arsalan Jawaid, Abdullah Karatas, Jörg Seewig |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 30 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-23006