Local regression on path spaces with signature metrics

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bayer, Christian, Gogolashvili, Davit, Pelizzari, Luca
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915563044864000
author Bayer, Christian
Gogolashvili, Davit
Pelizzari, Luca
author_facet Bayer, Christian
Gogolashvili, Davit
Pelizzari, Luca
contents We study nonparametric regression and classification for path-valued data. We introduce a functional Nadaraya-Watson estimator that combines the signature transform from rough path theory with local kernel regression. The signature transform provides a principled way to encode sequential data through iterated integrals, enabling direct comparison of paths in a natural metric space. Our approach leverages signature-induced distances within the classical kernel regression framework, achieving computational efficiency while avoiding the scalability bottlenecks of large-scale kernel matrix operations. We establish finite-sample convergence bounds demonstrating favorable statistical properties of signature-based distances compared to traditional metrics in infinite-dimensional settings. We propose robust signature variants that provide stability against outliers, enhancing practical performance. Applications to both synthetic and real-world data - including stochastic differential equation learning and time series classification - demonstrate competitive accuracy while offering significant computational advantages over existing methods.
format Preprint
id arxiv_https___arxiv_org_abs_2510_16728
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Local regression on path spaces with signature metrics
Bayer, Christian
Gogolashvili, Davit
Pelizzari, Luca
Machine Learning
Probability
Methodology
60L10, 60L20, 62G05, 62G08
We study nonparametric regression and classification for path-valued data. We introduce a functional Nadaraya-Watson estimator that combines the signature transform from rough path theory with local kernel regression. The signature transform provides a principled way to encode sequential data through iterated integrals, enabling direct comparison of paths in a natural metric space. Our approach leverages signature-induced distances within the classical kernel regression framework, achieving computational efficiency while avoiding the scalability bottlenecks of large-scale kernel matrix operations. We establish finite-sample convergence bounds demonstrating favorable statistical properties of signature-based distances compared to traditional metrics in infinite-dimensional settings. We propose robust signature variants that provide stability against outliers, enhancing practical performance. Applications to both synthetic and real-world data - including stochastic differential equation learning and time series classification - demonstrate competitive accuracy while offering significant computational advantages over existing methods.
title Local regression on path spaces with signature metrics
topic Machine Learning
Probability
Methodology
60L10, 60L20, 62G05, 62G08
url https://arxiv.org/abs/2510.16728