Noise-Robust Phase Connectivity Estimation via Bayesian Circular Functional Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Sugasawa, Shonosuke, Matsuda, Takeru, Nagakawa, Tomoyuki
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915484155248640
author Sugasawa, Shonosuke
Matsuda, Takeru
Nagakawa, Tomoyuki
author_facet Sugasawa, Shonosuke
Matsuda, Takeru
Nagakawa, Tomoyuki
contents The phase locking value (PLV) is a widely used measure to detect phase connectivity. Main drawbacks of the standard PLV are it can be sensitive to noisy observations and does not provide uncertainty measures under finite samples. To overcome the difficulty, we propose a model-based PLV through nonparametric statistical modeling. Specifically, since the discrete time series of phase can be regarded as a functional observation taking values on circle, we employ a Bayesian model for circular-variate functional data, which gives denoising and inference on the resulting PLV values. The proposed model is defined through "wrapping" functional Gaussian models on real line, for which we develop an efficient posterior computation algorithm using Gibbs sampler. The usefulness of the proposed method is demonstrated through simulation experiments based on real EEG data.
format Preprint
id arxiv_https___arxiv_org_abs_2509_06418
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Noise-Robust Phase Connectivity Estimation via Bayesian Circular Functional Models
Sugasawa, Shonosuke
Matsuda, Takeru
Nagakawa, Tomoyuki
Methodology
Applications
The phase locking value (PLV) is a widely used measure to detect phase connectivity. Main drawbacks of the standard PLV are it can be sensitive to noisy observations and does not provide uncertainty measures under finite samples. To overcome the difficulty, we propose a model-based PLV through nonparametric statistical modeling. Specifically, since the discrete time series of phase can be regarded as a functional observation taking values on circle, we employ a Bayesian model for circular-variate functional data, which gives denoising and inference on the resulting PLV values. The proposed model is defined through "wrapping" functional Gaussian models on real line, for which we develop an efficient posterior computation algorithm using Gibbs sampler. The usefulness of the proposed method is demonstrated through simulation experiments based on real EEG data.
title Noise-Robust Phase Connectivity Estimation via Bayesian Circular Functional Models
topic Methodology
Applications
url https://arxiv.org/abs/2509.06418