Single-trial effective brain connectivity patterns enhance discriminability of mental imagery tasks

Rathee, Dheeraj; Cecotti, Hubert; Prasad, Girijesh

doi:10.1088/1741-2552/aa785c

Cited by 36 publications

(34 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The non-stationarities of the EEG signals may be caused by various events, such as changes in the user attention levels, electrode placements, or user fatigues [11,12,13]. In other words, the basic cause of the non-stationarity in EEG signals is not only associated with the influences of the external stimuli to the brain mechanisms, but the switching of the cognitive task related inherent metastable states of neural assemblies also contributes towards it [14]. These non-stationarities cause notable variations or shifts in the EEG signals both during trial-to-trial, and session-to-session transfers [15,13,16,17].…”

Section: Introductionmentioning

confidence: 99%

Covariate shift estimation based adaptive ensemble learning for handling non-stationarity in motor imagery related EEG-based brain-computer interface

et al. 2019

Self Cite

View full text Add to dashboard Cite

The non-stationary nature of electroencephalography (EEG) signals makes an EEG-based brain-computer interface (BCI) a dynamic system, thus improving its performance is a challenging task. In addition, it is well-known that due to non-stationarity based covariate shifts, the input data distributions of EEG-based BCI systems change during inter-and intra-session transitions, which poses great difficulty for developments of online adaptive data-driven systems. Ensemble learning approaches have been used previously to tackle this challenge. However, passive scheme based implementation leads to poor efficiency while increasing high computational cost. This paper presents a novel integration of covariate shift estimation and unsupervised adaptive ensemble learning (CSE-UAEL) to tackle nonstationarity in motor-imagery (MI) related EEG classification. The proposed method first employs an exponentially weighted moving average model to detect the covariate shifts in the common spatial pattern features extracted from MI related brain responses. Then, a classifier ensemble was created and updated over time to account for changes in streaming input data distribution wherein new classifiers are added to the ensemble in accordance with estimated shifts. Furthermore, using two publicly available BCI-related EEG datasets, the proposed method was extensively compared with the state-of-the-art single-classifier based passive scheme, single-classifier based active scheme and ensemble based passive schemes. The experimental results show that the proposed active scheme based ensemble learning algorithm significantly enhances the BCI performance in MI classifications.

show abstract

Section: Introductionmentioning

confidence: 99%

Covariate shift estimation based adaptive ensemble learning for handling non-stationarity in motor imagery related EEG-based brain-computer interface

et al. 2019

Self Cite

View full text Add to dashboard Cite

show abstract

“…Firstly, time-domain potentials generated in response to specialized external stimulation (e.g., P300, steady-state evoked potentials, and evoked potentials), implemented in case of reactive BCI systems [4], [5]. Secondly, features generated from spontaneous brain signals generated during the performance of endogenous tasks, for instance, motor imagery (MI), emotion imagery, and mental arithmetic tasks, considered during active BCI paradigms [6], [7]. The latter approach of BCI implementation is highly popular, in particular, MI-related BCI which is one of the most explored EEG-based paradigms [8], [9], [10].…”

Section: Introductionmentioning

confidence: 99%

Current Source Density Estimation Enhances the Performance of Motor-Imagery-Related Brain–Computer Interface

Rathee

Raza

Prasad

et al. 2017

IEEE Trans. Neural Syst. Rehabil. Eng.

Self Cite

View full text Add to dashboard Cite

The objective is to evaluate the impact of EEG referencing schemes and spherical surface Laplacian (SSL) methods on the classification performance of motor-imagery (MI)-related brain-computer interface systems. Two EEG referencing schemes: common referencing and common average referencing and three surface Laplacian methods: current source density (CSD), finite difference method, and SSL using realistic head model were implemented separately for pre-processing of the EEG signals recorded at the scalp. A combination of filter bank common spatial filter for features extraction and support vector machine for classification was used for both pairwise binary classifications and four-class classification of MI tasks. The study provides three major outcomes: 1) the CSD method performs better than CR, providing a significant improvement of 3.02% and 5.59% across six binary classification tasks and four-class classification task, respectively; 2) the combination of a greater number of channels at the pre-processing stage as compared with the feature extraction stage yields better classification accuracies for all the Laplacian methods; and 3) the efficiency of all the surface Laplacian methods reduced significantly in the case of a fewer number of channels considered during the pre-processing.

show abstract

“…(P train (x) = P test (x)), while transitioning from the training to testing stage. A typical example of the CS for ten overlapping frequency bands ( [8][9][10][11][12], [10][11][12][13][14],... Hz) in the feature set of EEG data is illustrated in Figure 1.1 for the subject A07 of BCI competition-IV dataset 2A (the description of the dataset is present in section IV). For each plot, the blue solid ellipse and line show the input data distribution P train (x) and the classification hyperplane for training dataset, respectively.…”

Section: Covariate Shift In Eeg Signalsmentioning

confidence: 99%

“…A total of 10 band-pass filters (i.e. filter bank) with overlapping bandwidths, including [8][9][10][11][12], [10][11][12][13][14], [12][13][14][15][16], [14][15][16][17][18], [16][17][18][19][20], [18][19][20][21][22], [20][21][22][23][24], [22][23][24][25][26], [24][25][26][27][28], and [26][27][28][29][30] Hz were used for temporal filtering of the data. Next, spatial filtering using CSP algorithm was performed to maximize the dive...…”

Section: Signal Processing and Feature Extractionmentioning

confidence: 99%

“…The EEG signals are acquired through a multichannel EEG amplifier, and a pre-processing step is performed to reduce noise and enhances the signal-to-noise ratio. In the next step, the discriminable features are extracted from the artefact-cleaned signals using feature extraction techniques, such as spatial filtering (e.g., common spatial pattern (CSP), laplacian filtering, current source density) [7,8,9]. Further, the optimal features are employed for training the classifier model.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Covariate shift detection-based nonstationary adaptation in motor-imagery-based brain–computer interface

Raza

Rathee

Signal Processing and Machine Learning for Brain-Machine Interfaces

Self Cite

View full text Add to dashboard Cite

Non-stationary learning (NSL) refers to the process that can learn patterns from data, adapt to shifts, and improve performance of the system with its experience while operating in the non-stationary environments (NSEs). Covariate shift (CS) presents a major challenge during data processing within NSEs wherein the input-data distribution shifts during transitioning from training to testing phase. CS is one of the fundamental issues in electroencephalogram (EEG) based brain-computer interface (BCI) systems and can be often observed during multiple trials of EEG data recorded over different sessions. Thus, conventional learning algorithms struggle to accommodate these CSs in streaming EEG data resulting in low performance (in terms of classification accuracy) of motor imagery (MI)-related BCI systems. This chapter aims to introduce a novel framework for non-stationary adaptation in MI-related BCI system based on covariate shift detection (CSD) applied to the temporal and spatial filtered features extracted from raw EEG signals. The chapter collectively provides an efficient method for accounting non-stationarity in EEG data during learning in NSEs.

show abstract

Single-trial effective brain connectivity patterns enhance discriminability of mental imagery tasks

Abstract: We show that single-trial connectivity features for mixed imagery tasks (i.e. combination of CI and MI) can outperform the features obtained by current state-of-the-art method and hence can be successfully applied for BCI applications.

Cited by 36 publications

References 56 publications

Covariate shift estimation based adaptive ensemble learning for handling non-stationarity in motor imagery related EEG-based brain-computer interface

Covariate shift estimation based adaptive ensemble learning for handling non-stationarity in motor imagery related EEG-based brain-computer interface

Current Source Density Estimation Enhances the Performance of Motor-Imagery-Related Brain–Computer Interface

Covariate shift detection-based nonstationary adaptation in motor-imagery-based brain–computer interface

Contact Info

Product

Resources

About