Frequency Domain Reservoir Computing
Klaus Schertler, Xiomara Runge, Andrea Ceni, David Kappel, Claudio Gallicchio · 2026
Read Frequency Domain Reservoir Computing by Klaus Schertler, Xiomara Runge, Andrea Ceni, David Kappel, Claudio Gallicchio (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-2606-24969.txt.
Description
Klaus Schertler and four colleagues rebuild reservoir computing inside the frequency domain. Echo state networks train easily: recurrent weights stay fixed, so no backpropagation through time is needed and training has a closed form solution. Their weakness is scale. Expressive tasks demand large reservoirs, and updating a dense reservoir state costs quadratically in its size. The FRESCO architecture moves the whole computation into the frequency domain. A packed frequency domain readout, a dimensional zero padding input embedding, and a nonlinearity applied natively in that domain do the heavy lifting.
The result is linear complexity for dense nonlinear recurrent updates, with no domain shift overheads. Training and inference costs drop sharply. Performance does not suffer for the savings. FRESCO matches the state of the art on memory benchmarks, on sequence classification, and on multi series long horizon forecasting. Schertler and colleagues offer a scalable path for dense recurrent architectures.
Catalog record
| CREATOR | Klaus Schertler, Xiomara Runge, Andrea Ceni, David Kappel, Claudio Gallicchio |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 112 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.24969