Multiscale Granger causality analysis by à trous wavelet transform

Luca Faes, Daniele Marinazzo, Luca Faes, Iege Bassez, Sebastiano Stramaglia

Risultato della ricerca: Conference contribution

2 Citazioni (Scopus)

Abstract

Since interactions in neural systems occur across multiple temporal scales, it is likely that information flow will exhibit a multiscale structure, thus requiring a multiscale generalization of classical causality analysis like Granger's approach. However, the computation of multiscale measures of information dynamics is complicated by theoretical and practical issues such as filtering and undersampling: To overcome these problems, we propose a wavelet-based approach for multiscale Granger causality analysis, which is characterized by the following properties: (i) only the candidate driver variable is wavelet transformed (ii) the decomposition is performed using the à trous wavelet transform with cubic B-spline filter. We measure the causality, at a given scale, by including the wavelet coefficients of the driver times series, at that scale, in the regression model of the target. To validate our method, we apply it to public scalp EEG signals, and we find that the condition of closed eyes, at rest, is characterized by an enhanced Granger causality among channels at slow scales w.r.t. eye open condition, whilst the standard Granger causality is not significantly different in the two conditions.
Lingua originaleEnglish
Titolo della pubblicazione ospiteProceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017
Pagine25-28
Numero di pagine4
Stato di pubblicazionePublished - 2017

Fingerprint

Electroencephalography
wavelet analysis
Splines
Wavelet transforms
Time series
Decomposition
information flow
electroencephalography
splines
regression analysis
decomposition
filters
coefficients
interactions

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering
  • Instrumentation

Cita questo

Faes, L., Marinazzo, D., Faes, L., Bassez, I., & Stramaglia, S. (2017). Multiscale Granger causality analysis by à trous wavelet transform. In Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017 (pagg. 25-28)

Multiscale Granger causality analysis by à trous wavelet transform. / Faes, Luca; Marinazzo, Daniele; Faes, Luca; Bassez, Iege; Stramaglia, Sebastiano.

Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017. 2017. pag. 25-28.

Risultato della ricerca: Conference contribution

Faes, L, Marinazzo, D, Faes, L, Bassez, I & Stramaglia, S 2017, Multiscale Granger causality analysis by à trous wavelet transform. in Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017. pagg. 25-28.
Faes L, Marinazzo D, Faes L, Bassez I, Stramaglia S. Multiscale Granger causality analysis by à trous wavelet transform. In Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017. 2017. pag. 25-28
Faes, Luca ; Marinazzo, Daniele ; Faes, Luca ; Bassez, Iege ; Stramaglia, Sebastiano. / Multiscale Granger causality analysis by à trous wavelet transform. Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017. 2017. pagg. 25-28
@inproceedings{bb1a82662b6147c0b4d80a5e85312737,
title = "Multiscale Granger causality analysis by {\`a} trous wavelet transform",
abstract = "Since interactions in neural systems occur across multiple temporal scales, it is likely that information flow will exhibit a multiscale structure, thus requiring a multiscale generalization of classical causality analysis like Granger's approach. However, the computation of multiscale measures of information dynamics is complicated by theoretical and practical issues such as filtering and undersampling: To overcome these problems, we propose a wavelet-based approach for multiscale Granger causality analysis, which is characterized by the following properties: (i) only the candidate driver variable is wavelet transformed (ii) the decomposition is performed using the {\~A}  trous wavelet transform with cubic B-spline filter. We measure the causality, at a given scale, by including the wavelet coefficients of the driver times series, at that scale, in the regression model of the target. To validate our method, we apply it to public scalp EEG signals, and we find that the condition of closed eyes, at rest, is characterized by an enhanced Granger causality among channels at slow scales w.r.t. eye open condition, whilst the standard Granger causality is not significantly different in the two conditions.",
author = "Luca Faes and Daniele Marinazzo and Luca Faes and Iege Bassez and Sebastiano Stramaglia",
year = "2017",
language = "English",
isbn = "9781509067060; 978-1-5090-6707-7",
pages = "25--28",
booktitle = "Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017",

}

TY - GEN

T1 - Multiscale Granger causality analysis by à trous wavelet transform

AU - Faes, Luca

AU - Marinazzo, Daniele

AU - Faes, Luca

AU - Bassez, Iege

AU - Stramaglia, Sebastiano

PY - 2017

Y1 - 2017

N2 - Since interactions in neural systems occur across multiple temporal scales, it is likely that information flow will exhibit a multiscale structure, thus requiring a multiscale generalization of classical causality analysis like Granger's approach. However, the computation of multiscale measures of information dynamics is complicated by theoretical and practical issues such as filtering and undersampling: To overcome these problems, we propose a wavelet-based approach for multiscale Granger causality analysis, which is characterized by the following properties: (i) only the candidate driver variable is wavelet transformed (ii) the decomposition is performed using the à trous wavelet transform with cubic B-spline filter. We measure the causality, at a given scale, by including the wavelet coefficients of the driver times series, at that scale, in the regression model of the target. To validate our method, we apply it to public scalp EEG signals, and we find that the condition of closed eyes, at rest, is characterized by an enhanced Granger causality among channels at slow scales w.r.t. eye open condition, whilst the standard Granger causality is not significantly different in the two conditions.

AB - Since interactions in neural systems occur across multiple temporal scales, it is likely that information flow will exhibit a multiscale structure, thus requiring a multiscale generalization of classical causality analysis like Granger's approach. However, the computation of multiscale measures of information dynamics is complicated by theoretical and practical issues such as filtering and undersampling: To overcome these problems, we propose a wavelet-based approach for multiscale Granger causality analysis, which is characterized by the following properties: (i) only the candidate driver variable is wavelet transformed (ii) the decomposition is performed using the à trous wavelet transform with cubic B-spline filter. We measure the causality, at a given scale, by including the wavelet coefficients of the driver times series, at that scale, in the regression model of the target. To validate our method, we apply it to public scalp EEG signals, and we find that the condition of closed eyes, at rest, is characterized by an enhanced Granger causality among channels at slow scales w.r.t. eye open condition, whilst the standard Granger causality is not significantly different in the two conditions.

UR - http://hdl.handle.net/10447/273036

M3 - Conference contribution

SN - 9781509067060; 978-1-5090-6707-7

SP - 25

EP - 28

BT - Proceedings - 7th International Workshop on Advances in Sensors and Interfaces, IWASI 2017

ER -