Tangent space features-based transfer learning classification model for two-class motor imagery brain-computer interface

Pramod Gaur, Karl McCreadie, R. B. Pachori, H. Wang, Girijesh Prasad

Research output: Contribution to journalArticle

Abstract

The performance of a brain-computer interface (BCI) will generally improve by increasing the volume of training data on which it is trained. However, a classifier's generalization ability is often negatively affected when highly non-stationary data are collected across both sessions and subjects. The aim of this work is to reduce the long calibration time in BCI systems by proposing a transfer learning model which can be used for evaluating unseen single trials for a subject without the need for training session data. A method is proposed which combines a generalization of the previously proposed subject-specific "multivariate empirical-mode decomposition" preprocessing technique by taking a fixed band of 8-30Hz for all four motor imagery tasks and a novel classification model which exploits the structure of tangent space features drawn from the Riemannian geometry framework, that is shared among the training data of multiple sessions and subjects. Results demonstrate comparable performance improvement across multiple subjects without subject-specific calibration, when compared with other state-of-the-art techniques.

LanguageEnglish
Article number1950025
JournalInternational Journal of Neural Systems (IJNS)
Volume29
Issue number10
Early online date12 Nov 2019
DOIs
Publication statusPublished - 1 Dec 2019

Fingerprint

Brain computer interface
Calibration
Classifiers
Decomposition
Geometry

Keywords

  • Motor imagery
  • brain–computer interface (BCI)
  • tangent space
  • covariance matrix
  • multivariate empirical-mode decomposition (MEMD)
  • subject-specific multivariate empirical-mode decomposition-based filtering (SS-MEMDBF)
  • brain-computer interface (BCI)

Cite this

@article{668c78717b0043ec9b15e7f04782e472,
title = "Tangent space features-based transfer learning classification model for two-class motor imagery brain-computer interface",
abstract = "The performance of a brain-computer interface (BCI) will generally improve by increasing the volume of training data on which it is trained. However, a classifier's generalization ability is often negatively affected when highly non-stationary data are collected across both sessions and subjects. The aim of this work is to reduce the long calibration time in BCI systems by proposing a transfer learning model which can be used for evaluating unseen single trials for a subject without the need for training session data. A method is proposed which combines a generalization of the previously proposed subject-specific {"}multivariate empirical-mode decomposition{"} preprocessing technique by taking a fixed band of 8-30Hz for all four motor imagery tasks and a novel classification model which exploits the structure of tangent space features drawn from the Riemannian geometry framework, that is shared among the training data of multiple sessions and subjects. Results demonstrate comparable performance improvement across multiple subjects without subject-specific calibration, when compared with other state-of-the-art techniques.",
keywords = "Motor imagery, brain–computer interface (BCI), tangent space, covariance matrix, multivariate empirical-mode decomposition (MEMD), subject-specific multivariate empirical-mode decomposition-based filtering (SS-MEMDBF), brain-computer interface (BCI)",
author = "Pramod Gaur and Karl McCreadie and Pachori, {R. B.} and H. Wang and Girijesh Prasad",
year = "2019",
month = "12",
day = "1",
doi = "10.1142/S0129065719500254",
language = "English",
volume = "29",
journal = "International Journal of Neural Systems",
issn = "0129-0657",
number = "10",

}

TY - JOUR

T1 - Tangent space features-based transfer learning classification model for two-class motor imagery brain-computer interface

AU - Gaur, Pramod

AU - McCreadie, Karl

AU - Pachori, R. B.

AU - Wang, H.

AU - Prasad, Girijesh

PY - 2019/12/1

Y1 - 2019/12/1

N2 - The performance of a brain-computer interface (BCI) will generally improve by increasing the volume of training data on which it is trained. However, a classifier's generalization ability is often negatively affected when highly non-stationary data are collected across both sessions and subjects. The aim of this work is to reduce the long calibration time in BCI systems by proposing a transfer learning model which can be used for evaluating unseen single trials for a subject without the need for training session data. A method is proposed which combines a generalization of the previously proposed subject-specific "multivariate empirical-mode decomposition" preprocessing technique by taking a fixed band of 8-30Hz for all four motor imagery tasks and a novel classification model which exploits the structure of tangent space features drawn from the Riemannian geometry framework, that is shared among the training data of multiple sessions and subjects. Results demonstrate comparable performance improvement across multiple subjects without subject-specific calibration, when compared with other state-of-the-art techniques.

AB - The performance of a brain-computer interface (BCI) will generally improve by increasing the volume of training data on which it is trained. However, a classifier's generalization ability is often negatively affected when highly non-stationary data are collected across both sessions and subjects. The aim of this work is to reduce the long calibration time in BCI systems by proposing a transfer learning model which can be used for evaluating unseen single trials for a subject without the need for training session data. A method is proposed which combines a generalization of the previously proposed subject-specific "multivariate empirical-mode decomposition" preprocessing technique by taking a fixed band of 8-30Hz for all four motor imagery tasks and a novel classification model which exploits the structure of tangent space features drawn from the Riemannian geometry framework, that is shared among the training data of multiple sessions and subjects. Results demonstrate comparable performance improvement across multiple subjects without subject-specific calibration, when compared with other state-of-the-art techniques.

KW - Motor imagery

KW - brain–computer interface (BCI)

KW - tangent space

KW - covariance matrix

KW - multivariate empirical-mode decomposition (MEMD)

KW - subject-specific multivariate empirical-mode decomposition-based filtering (SS-MEMDBF)

KW - brain-computer interface (BCI)

UR - http://www.scopus.com/inward/record.url?scp=85074994960&partnerID=8YFLogxK

U2 - 10.1142/S0129065719500254

DO - 10.1142/S0129065719500254

M3 - Article

VL - 29

JO - International Journal of Neural Systems

T2 - International Journal of Neural Systems

JF - International Journal of Neural Systems

SN - 0129-0657

IS - 10

M1 - 1950025

ER -