Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2412.09800 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929723430404096 |
|---|---|
| author | Grigoryeva, Lyudmila Ting, Hannah Lim Jing Ortega, Juan-Pablo |
| author_facet | Grigoryeva, Lyudmila Ting, Hannah Lim Jing Ortega, Juan-Pablo |
| contents | Next-generation reservoir computing (NG-RC) has attracted much attention due to its excellent performance in spatio-temporal forecasting of complex systems and its ease of implementation. This paper shows that NG-RC can be encoded as a kernel ridge regression that makes training efficient and feasible even when the space of chosen polynomial features is very large. Additionally, an extension to an infinite number of covariates is possible, which makes the methodology agnostic with respect to the lags into the past that are considered as explanatory factors, as well as with respect to the number of polynomial covariates, an important hyperparameter in traditional NG-RC. We show that this approach has solid theoretical backing and good behavior based on kernel universality properties previously established in the literature. Various numerical illustrations show that these generalizations of NG-RC outperform the traditional approach in several forecasting applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_09800 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Infinite-dimensional next-generation reservoir computing Grigoryeva, Lyudmila Ting, Hannah Lim Jing Ortega, Juan-Pablo Machine Learning Neural and Evolutionary Computing Computational Physics Next-generation reservoir computing (NG-RC) has attracted much attention due to its excellent performance in spatio-temporal forecasting of complex systems and its ease of implementation. This paper shows that NG-RC can be encoded as a kernel ridge regression that makes training efficient and feasible even when the space of chosen polynomial features is very large. Additionally, an extension to an infinite number of covariates is possible, which makes the methodology agnostic with respect to the lags into the past that are considered as explanatory factors, as well as with respect to the number of polynomial covariates, an important hyperparameter in traditional NG-RC. We show that this approach has solid theoretical backing and good behavior based on kernel universality properties previously established in the literature. Various numerical illustrations show that these generalizations of NG-RC outperform the traditional approach in several forecasting applications. |
| title | Infinite-dimensional next-generation reservoir computing |
| topic | Machine Learning Neural and Evolutionary Computing Computational Physics |
| url | https://arxiv.org/abs/2412.09800 |