Learning dynamical systems with hit-and-run random feature maps

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Mandal, Pinak, Gottwald, Georg A.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917972327530496
author Mandal, Pinak
Gottwald, Georg A.
author_facet Mandal, Pinak
Gottwald, Georg A.
contents We show how random feature maps can be used to forecast dynamical systems with excellent forecasting skill. We consider the tanh activation function and judiciously choose the internal weights in a data-driven manner such that the resulting features explore the nonlinear, non-saturated regions of the activation function. We introduce skip connections and construct a deep variant of random feature maps by combining several units. To mitigate the curse of dimensionality, we introduce localization where we learn local maps, employing conditional independence. Our modified random feature maps provide excellent forecasting skill for both single trajectory forecasts as well as long-time estimates of statistical properties, for a range of chaotic dynamical systems with dimensions up to 512. In contrast to other methods such as reservoir computers which require extensive hyperparameter tuning, we effectively need to tune only a single hyperparameter, and are able to achieve state-of-the-art forecast skill with much smaller networks.
format Preprint
id arxiv_https___arxiv_org_abs_2501_06661
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning dynamical systems with hit-and-run random feature maps
Mandal, Pinak
Gottwald, Georg A.
Machine Learning
Data Analysis, Statistics and Probability
Methodology
We show how random feature maps can be used to forecast dynamical systems with excellent forecasting skill. We consider the tanh activation function and judiciously choose the internal weights in a data-driven manner such that the resulting features explore the nonlinear, non-saturated regions of the activation function. We introduce skip connections and construct a deep variant of random feature maps by combining several units. To mitigate the curse of dimensionality, we introduce localization where we learn local maps, employing conditional independence. Our modified random feature maps provide excellent forecasting skill for both single trajectory forecasts as well as long-time estimates of statistical properties, for a range of chaotic dynamical systems with dimensions up to 512. In contrast to other methods such as reservoir computers which require extensive hyperparameter tuning, we effectively need to tune only a single hyperparameter, and are able to achieve state-of-the-art forecast skill with much smaller networks.
title Learning dynamical systems with hit-and-run random feature maps
topic Machine Learning
Data Analysis, Statistics and Probability
Methodology
url https://arxiv.org/abs/2501.06661