Motor Imagery Electroencephalogram Research Articles

Multivariate variational mode decomposition & phase space reconstruction based motor imagery EEG classification

Electroencephalogram signals have gained much attention in recent years as these signals help in decoding a person's intentions for performing various actions using the Brain Computer Interface system. However, it is challenging to accurately classify human intentions as the electroencephalogram signals are non-linear and non-stationary. This work proposes a novel methodology for feature extraction and classification of motor imagery electroencephalogram signals. The proposed methodology is based on Multivariate Variational Mode Decomposition and Phase Space Reconstruction. A number of statistical and non-linear features are extracted and classified using the Ensemble Support Vector Machine model. The efficiency of the methodology is evaluated on BCI competition-IV(2a) dataset. The evaluation results conclude that the proposed methodology is able to classify both binary and multi-class electroencephalogram data with an average classification accuracy of 96.76% and 84.22%, respectively. Hence, the proposed methodology can be utilized in designing a real-time Brain Computer interface system.

Motor Imagery Multi-Tasks Classification for BCIs Using the NVIDIA Jetson TX2 Board and the EEGNet Network.

Nowadays, Brain-Computer Interfaces (BCIs) still captivate large interest because of multiple advantages offered in numerous domains, explicitly assisting people with motor disabilities in communicating with the surrounding environment. However, challenges of portability, instantaneous processing time, and accurate data processing remain for numerous BCI system setups. This work implements an embedded multi-tasks classifier based on motor imagery using the EEGNet network integrated into the NVIDIA Jetson TX2 card. Therefore, two strategies are developed to select the most discriminant channels. The former uses the accuracy based-classifier criterion, while the latter evaluates electrode mutual information to form discriminant channel subsets. Next, the EEGNet network is implemented to classify discriminant channel signals. Additionally, a cyclic learning algorithm is implemented at the software level to accelerate the model learning convergence and fully profit from the NJT2 hardware resources. Finally, motor imagery Electroencephalogram (EEG) signals provided by HaLT's public benchmark were used, in addition to the k-fold cross-validation method. Average accuracies of 83.7% and 81.3% were achieved by classifying EEG signals per subject and motor imagery task, respectively. Each task was processed with an average latency of 48.7 ms. This framework offers an alternative for online EEG-BCI systems' requirements, dealing with short processing times and reliable classification accuracy.

Motor imagery classification method based on relative wavelet packet entropy brain network and improved lasso.

Motor imagery (MI) electroencephalogram (EEG) signals have a low signal-to-noise ratio, which brings challenges in feature extraction and feature selection with high classification accuracy. In this study, we proposed an approach that combined an improved lasso with relief-f to extract the wavelet packet entropy features and the topological features of the brain function network. For signal denoising and channel filtering, raw MI EEG was filtered based on an R2 map, and then the wavelet soft threshold and one-to-one multi-class score common spatial pattern algorithms were used. Subsequently, the relative wavelet packet entropy and corresponding topological features of the brain network were extracted. After feature fusion, mutcorLasso and the relief-f method were applied for feature selection, followed by three classifiers and an ensemble classifier, respectively. The experiments were conducted on two public EEG datasets (BCI Competition III dataset IIIa and BCI Competition IV dataset IIa) to verify this proposed method. The results showed that the brain network topology features and feature selection methods can retain the information of EEG more effectively and reduce the computational complexity, and the average classification accuracy for both public datasets was above 90%; hence, this algorithms is suitable in MI-BCI and has potential applications in rehabilitation and other fields.

A Cross-Space CNN With Customized Characteristics for Motor Imagery EEG Classification

The classification of motor imagery-electroencephalogram(MI-EEG)based brain-computer interface(BCI)can be used to decode neurological activities, which has been widely applied in the control of external devices. However, two factors still hinder the improvement of classification accuracy and robustness, especially in multi-class tasks. First, existing algorithms are based on a single space (measuring or source space). They suffer from the holistic low spatial resolution of the measuring space or the locally high spatial resolution information accessed from the source space, failing to provide holistic and high-resolution representations. Second, the subject specificity is not sufficiently characterized, resulting in the loss of personalized intrinsic information. Therefore, we propose a cross-space convolutional neural network (CS-CNN) with customized characteristics for four-class MI-EEG classification. This algorithm uses the modified customized band common spatial patterns (CBCSP) and duplex mean-shift clustering (DMSClustering) to express the specific rhythms and source distribution information in cross-space. At the same time, multi-view features from the time, frequency and space domains are extracted, connecting with CNN to fuse the characteristics from two spaces and classify them. MI-EEG was collected from 20 subjects. Lastly, the classification accuracy of the proposed is 96.05% with real MRI information and 94.79% without MRI in the private dataset. And the results in the BCI competition IV-2a show that CS-CNN outperforms the state-of-the-art algorithms, achieving an accuracy improvement of 1.98%, and a standard deviation reduction of 5.15%.

Enhancement of time-frequency energy for the classification of motor imagery electroencephalogram based on an improved FitzHugh-Nagumo neuron system.

Brain-computer interface (BCI) based on motor imagery (MI) electroencephalogram (EEG) has become an essential way for rehabilitation, because of the activation and interaction of motor neurons between the brain and rehabilitation devices in recent years. However, due to the discrepancies between individuals, the frequency ranges can be different even for the same rhythm component of EEG recordings, which brings difficulties to the extraction of features for MI classification. Typical algorithms for MI classification such as common spatial patterns (CSP) require multi-channel analysis and lack frequency information. With the development of BCI, the single-channel BCI system has become indispensable for simplicity of use. However, the currently available single-channel detection methods have low classification accuracy. To address this issue, two novel frameworks based on an improved two-dimensional nonlinear FitzHugh-Nagumo (FHN) neuron system are proposed to extract features of the single-channel MI. To evaluate the effectiveness of the proposed methods, this research utilized an open-access database (BCI competition IV dataset 2a), an offline database, and a 10-fold cross-validation procedure. Experimental results showed that the improved nonlinear FHN system can transfer the energy of noise into MI, thereby effectively enhancing the time-frequency energy. Compared with the traditional methods, the proposed methods can achieve higher classification accuracy and robustness.

Joint Filter-Band-Combination and Multi-view CNN for Electroencephalogram Decoding.

Motor imagery (MI) electroencephalogram (EEG) signals have an important role in brain-computer interface (BCI) research. However, effectively decoding these signals remains a problem to be solved. Traditional EEG signal decoding algorithms rely on parameter design to extract features, whereas deep learning algorithms represented by convolution neural network (CNN) can automatically extract features, which is more suitable for BCI applications. However, when EEG data is taken as input in raw time series, traditional 1D-CNNs are unable to acquire both frequency domain and channel association information. To solve this problem, this study proposes a novel algorithm by inserting two modules into CNN. One is the Filter Band Combination (FBC) Module, which preserves as many frequency domain features as possible while maintaining the time domain characteristics of EEG. Another module is Multi-View structure that can extract features from the output of FBC module. To prevent over fitting, we used a cosine annealing algorithm with restart strategy to update the learning rate. The proposed algorithm was validated on the BCI competition dataset and the experiment dataset, using accuracy, standard deviation, and kappa coefficient. Compared with traditional decoding algorithms, our proposed algorithm achieved an improvement of the maximum average correct rate of 6.6% on the motion imagery 4-classes recognition mission and 11.3% on the 2-classes classification task.

An Adaptive Channel Selection and Graph ResNet Based Algorithm for Motor Imagery Classification

In Brain-Computer interface (BCI) applications, achieving accurate control relies heavily on the classification accuracy and efficiency of motor imagery electroencephalogram (EEG) signals. However, factors such as mutual interference between multi-channel signals, inter-individual variability, and noise interference in the channels pose challenges to motor imagery EEG signal classification. To address these problems, this paper proposes an Adaptive Channel Selection algorithm aimed at optimizing classification accuracy and Information Translate Rate (ITR). First, C3, C4, and Cz are selected as key channels based on neurophysiological evidence and extensive experimental studies. Next, the channel selection is fine-tuned using spatial location and absolute Pearson correlation coefficients. By analyzing the relationship between EEG channels and key channels, the most relevant channel combination is determined for each subject, reducing confounding information and improving classification accuracy. To validate the method, the SHU Dataset and the PhysioNet Dataset are used in experiments. The Graph ResNet classification model is employed to extract features from the selected channel combinations using deep learning techniques. Experimental results show that the average classification accuracy is improved by 5.36% and 9.19%, and the Information Translate Rate is improved by 29.24% and 26.75%, respectively, compared to a single channel combination.

Research on the feature representation of motor imagery electroencephalogram signal based on individual adaptation

Aiming at the problem of low recognition accuracy of motor imagery electroencephalogram signal due to individual differences of subjects, an individual adaptive feature representation method of motor imagery electroencephalogram signal is proposed in this paper. Firstly, based on the individual differences and signal characteristics in different frequency bands, an adaptive channel selection method based on expansive relevant features with label F (ReliefF) was proposed. By extracting five time-frequency domain observation features of each frequency band signal, ReliefF algorithm was employed to evaluate the effectiveness of the frequency band signal in each channel, and then the corresponding signal channel was selected for each frequency band. Secondly, a feature representation method of common space pattern (CSP) based on fast correlation-based filter (FCBF) was proposed (CSP-FCBF). The features of electroencephalogram signal were extracted by CSP, and the best feature sets were obtained by using FCBF to optimize the features, so as to realize the effective state representation of motor imagery electroencephalogram signal. Finally, support vector machine (SVM) was adopted as a classifier to realize identification. Experimental results show that the proposed method in this research can effectively represent the states of motor imagery electroencephalogram signal, with an average identification accuracy of (83.0±5.5)% for four types of states, which is 6.6% higher than the traditional CSP feature representation method. The research results obtained in the feature representation of motor imagery electroencephalogram signal lay the foundation for the realization of adaptive electroencephalogram signal decoding and its application.

Multi-task motor imagery electroencephalogram classification based on adaptive time-frequency common spatial pattern combined with convolutional neural network

The effective classification of multi-task motor imagery electroencephalogram (EEG) is helpful to achieve accurate multi-dimensional human-computer interaction, and the high frequency domain specificity between subjects can improve the classification accuracy and robustness. Therefore, this paper proposed a multi-task EEG signal classification method based on adaptive time-frequency common spatial pattern (CSP) combined with convolutional neural network (CNN). The characteristics of subjects' personalized rhythm were extracted by adaptive spectrum awareness, and the spatial characteristics were calculated by using the one-versus-rest CSP, and then the composite time-domain characteristics were characterized to construct the spatial-temporal frequency multi-level fusion features. Finally, the CNN was used to perform high-precision and high-robust four-task classification. The algorithm in this paper was verified by the self-test dataset containing 10 subjects (33 ± 3 years old, inexperienced) and the dataset of the 4th 2018 Brain-Computer Interface Competition (BCI competition Ⅳ-2a). The average accuracy of the proposed algorithm for the four-task classification reached 93.96% and 84.04%, respectively. Compared with other advanced algorithms, the average classification accuracy of the proposed algorithm was significantly improved, and the accuracy range error between subjects was significantly reduced in the public dataset. The results show that the proposed algorithm has good performance in multi-task classification, and can effectively improve the classification accuracy and robustness.

Classification of motor imagery electroencephalogram signals by using adaptive cross-subject transfer learning.

Electroencephalogram (EEG)-based motor imagery (MI) classification is an important aspect in brain-computer interfaces (BCIs), which bridges between neural system and computer devices decoding brain signals into recognizable machine commands. However, due to the small number of training samples of MI electroencephalogram (MI-EEG) for a single subject and the great individual differences of MI-EEG among different subjects, the generalization and accuracy of the model on the specific MI task may be poor. To solve these problems, an adaptive cross-subject transfer learning algorithm is proposed, which is based on kernel mean matching (KMM) and transfer learning adaptive boosting (TrAdaBoost) method. First, the common spatial pattern (CSP) is used to extract the spatial features. Then, in order to make the feature distribution more similar among different subjects, the KMM algorithm is used to compute a sample weight matrix for aligning the mean between source and target domains and reducing distribution differences among different subjects. Finally, the sample weight matrix from KMM is used as the initialization weight of TrAdaBoost, and then TrAdaBoost is used to adaptively select source domain samples that are closer to the target task distribution to assist in building a classification model. In order to verify the effectiveness and feasibility of the proposed method, the algorithm is applied to BCI Competition IV datasets and in-house datasets. The results show that the average classification accuracy of the proposed method on the public datasets is 89.1%, and the average classification accuracy on the in-house datasets is 80.4%. Compared with the existing methods, the proposed method effectively improves the classification accuracy of MI-EEG signals. At the same time, this paper also applies the proposed algorithm to the in-house dataset, the results verify the effectiveness of the algorithm again, and the results of this study have certain clinical guiding significance for brain rehabilitation.

Feature optimization based on improved novel global harmony search algorithm for motor imagery electroencephalogram classification.

Effectively decoding electroencephalogram (EEG) pattern for specific mental tasks is a crucial topic in the development of brain-computer interface (BCI). Extracting common spatial pattern (CSP) features from motor imagery EEG signals is often highly dependent on the selection of frequency band and time interval. Therefore, optimizing frequency band and time interval would contribute to effective feature extraction and accurate EEG decoding. This study proposes an approach based on an improved novel global harmony search (INGHS) to optimize frequency-time parameters for effective CSP feature extraction. The INGHS algorithm is applied to find the optimal frequency band and temporal interval. The linear discriminant analysis and support vector machine are used for EEG pattern decoding. Extensive experimental studies are conducted on three EEG datasets to assess the effectiveness of our proposed method. The average test accuracy obtained by the time-frequency parameters selected by the proposed INGHS method is slightly better than artificial bee colony (ABC) and particle swarm optimization (PSO) algorithms. Furthermore, the INGHS algorithm is superior to PSO and ABC in running time. These superior experimental results demonstrate that the optimal frequency band and time interval selected by the INGHS algorithm could significantly improve the decoding accuracy compared with the traditional CSP method. This method has a potential to improve the performance of MI-based BCI systems.

Learning Optimal Time-Frequency-Spatial Features by the CiSSA-CSP Method for Motor Imagery EEG Classification.

The common spatial pattern (CSP) is a popular method in feature extraction for motor imagery (MI) electroencephalogram (EEG) classification in brain-computer interface (BCI) systems. However, combining temporal and spectral information in the CSP-based spatial features is still a challenging issue, which greatly affects the performance of MI-based BCI systems. Here, we propose a novel circulant singular spectrum analysis embedded CSP (CiSSA-CSP) method for learning the optimal time-frequency-spatial features to improve the MI classification accuracy. Specifically, raw EEG data are first segmented into multiple time segments and spectrum-specific sub-bands are further derived by CiSSA from each time segment in a set of non-overlapping filter bands. CSP features extracted from all time-frequency segments contain more sufficient time-frequency-spatial information. An experimental study was implemented on the publicly available EEG dataset (BCI Competition III dataset IVa) and a self-collected experimental EEG dataset to validate the effectiveness of the CiSSA-CSP method. Experimental results demonstrate that discriminative and robust features are extracted effectively. Compared with several state-of-the-art methods, the proposed method exhibited optimal accuracies of 96.6% and 95.2% on the public and experimental datasets, respectively, which confirms that it is a promising method for improving the performance of MI-based BCIs.

3D Convolution neural network with multiscale spatial and temporal cues for motor imagery EEG classification.

Recently, deep learning-based methods have achieved meaningful results in the Motor imagery electroencephalogram (MI EEG) classification. However, because of the low signal-to-noise ratio and the various characteristics of brain activities among subjects, these methods lack a subject adaptive feature extraction mechanism. Another issue is that they neglect important spatial topological information and the global temporal variation trend of MI EEG signals. These issues limit the classification accuracy. Here, we propose an end-to-end 3D CNN to extract multiscale spatial and temporal dependent features for improving the accuracy performance of 4-class MI EEG classification. The proposed method adaptively assigns higher weights to motor-related spatial channels and temporal sampling cues than the motor-unrelated ones across all brain regions, which can prevent influences caused by biological and environmental artifacts. Experimental evaluation reveals that the proposed method achieved an average classification accuracy of 93.06% and 97.05% on two commonly used datasets, demonstrating excellent performance and robustness for different subjects compared to other state-of-the-art methods.In order to verify the real-time performance in actual applications, the proposed method is applied to control the robot based on MI EEG signals. The proposed approach effectively addresses the issues of existing methods, improves the classification accuracy and the performance of BCI system, and has great application prospects.

Optimal channel and frequency band‐based feature selection for motor imagery electroencephalogram classification

AbstractCommon spatial pattern (CSP) is a widely adopted method for electroencephalogram (EEG) feature extraction in brain‐computer interface (BCI) based on motor imagery. Bandpass‐filtering EEG into several subbands related to brain activity tasks is an effective approach to improve the performance of CSP based algorithm. However, this approach tends to suffer the over‐fitting problem because of the increase in feature dimension. Therefore, we proposed an optimal channel and frequency band‐based CSP feature selection method in this paper. Firstly, the correlation coefficient was calculated to select the optimal channels, and these channels were bandpass‐filtered into multiple overlapping subbands. The subbands with higher power spectrum density were chosen for CSP feature extraction. Next, the pair‐wise relevance was utilized to remove subband features with less difference. And then the screened subband features were combined with features extracted from the broadband signal. The Fisher ratio was exploited to carry out further feature selection. Finally, a support vector machine (SVM) was trained to classify the selected CSP features. An experimental study was implemented on BCI competition III dataset IVa and BCI competition IV dataset 1. The average classification accuracy reached 89.33% and 84.08%, which indicated the rationality and effectiveness of the proposed method.

Motor Imagery BCI Classification Based on Multivariate Variational Mode Decomposition

In this article, a novel computer-aided diagnosis framework is proposed for the classification of motor imagery (MI) electroencephalogram (EEG) signals. First, a multivariate variational mode decomposition (MVMD) method was employed to obtain joint modes in frequency scale across all channels. Second, several multi-domain features (time domain, frequency domain, nonlinear and geometrical) were extracted from each EEG signal, and to further enhance the classification performance of different MI EEG signals, a variety of wrapper and filter feature selection methods were utilized with different channel combinations. Finally, to avoid a large number of training sessions for a particular device, extensive subject-independent experiments were performed. The MVMD applied to 18-channel EEG from the motor cortex area in combination with the ReliefF feature selection method achieved an average classification accuracy of 99.8% for a subject-dependent while 98.3% for subject-independent experiments. Besides the aforementioned combination provide above 99% accuracy for subjects with sufficient or small training samples for both subject-dependent or independent cases. These promising findings suggest that the proposed framework is flexible to use for subject-dependent or independent BCI systems.

A multi-classification algorithm based on multi-domain information fusion for motor imagery BCI

The current problem of motor imagery Electroencephalogram(EEG) signal classification is low classification accuracy and fixed EEG channel selection. We proposed a novel classification algorithm for motor imagery EEG signals, which overcomes the contradiction between the number of channels and the representational ability of features. Higher classification accuracy is achieved using less number of channels. The algorithm makes a combination of time windows, filter banks, and an optimal sorting of the projection space to reveal multi-domain information. Experiments based on the two datasets of BCI Competition have proved that the channel selection strategy used in this paper can adapt to the subject’s neural information and select the optimal channel combination. The feature extraction algorithm proposed can achieve excellent classification accuracy (77.7 %) and kappa value (0.70). The results are improved by 26.2 % compared to the One Versus One-Common Spatial Pattern (OVO-CSP) method and by 8.2 % compared to the One Versus One-Filter bank common spatial pattern (OVO-FBCSP) method. Additionally, the proposed method has outperformed to the other state-of-the-art methods using the same data set in terms of the performance. The proposed methodology can be employed as a promising tool for a motor imagery BCI device.

A dual alignment-based multi-source domain adaptation framework for motor imagery EEG classification.

Domain adaptation, as an important branch of transfer learning, can be applied to cope with data insufficiency and high subject variabilities in motor imagery electroencephalogram (MI-EEG) based brain-computer interfaces. The existing methods generally focus on aligning data and feature distribution; however, aligning each source domain with the informative samples of the target domain and seeking the most appropriate source domains to enhance the classification effect has not been considered. In this paper, we propose a dual alignment-based multi-source domain adaptation framework, denoted DAMSDAF. Based on continuous wavelet transform, all channels of MI-EEG signals are converted respectively and the generated time-frequency spectrum images are stitched to construct multi-source domains and target domain. Then, the informative samples close to the decision boundary are found in the target domain by using entropy, and they are employed to align and reassign each source domain with normalized mutual information. Furthermore, a multi-branch deep network (MBDN) is designed, and the maximum mean discrepancy is embedded in each branch to realign the specific feature distribution. Each branch is separately trained by an aligned source domain, and all the single branch transfer accuracies are arranged in descending order and utilized for weighted prediction of MBDN. Therefore, the most suitable number of source domains with top weights can be automatically determined. Extensive experiments are conducted based on 3 public MI-EEG datasets. DAMSDAF achieves the classification accuracies of 92.56%, 69.45% and 89.57%, and the statistical analysis is performed by the kappa value and t-test. Experimental results show that DAMSDAF significantly improves the transfer effects compared to the present methods, indicating that dual alignment can sufficiently use the different weighted samples and even source domains at different levels as well as realizing optimal selection of multi-source domains.

Excellent fine-tuning: From specific-subject classification to cross-task classification for motor imagery

With the popularity of deep learning, motor imagery electroencephalogram (MI-EEG) recognition based on feature extractors and classifiers has performed well. However, the features extracted by most models are not discriminative enough and are limited to specific-subject classifi-cation. We proposed a novel model Metric-based Spatial Filtering Transformer (MSFT) that utilizes additive angular margin loss to enforce the deep model to improve inter-class separability while enhancing intra-class compactness. Besides, a data augmentation method called EEG pyramid was applied to the model. Our model not only outperforms many recent benchmarks in specific-subject classifi-cation, but also is used for cross-subject and even cross-task classification. We did some experiments using BCI competition IV 2a and 2b datasets to evaluate the average accuracy. The Specific-subject: 86.11 % for 2a, 88.39 % for 2b. The Cross-subject: 61.92 % for 2a. The Cross-task: training the feature extractor with 2a data and then fine-tuning the classifier with 2b can achieve an average accuracy of 83.38 %. Our method is more general than most benchmarks and can deal with different kinds of classification situations.

A Fused Multidimensional EEG Classification Method Based on an Extreme Tree Feature Selection.

When a brain-computer interface (BCI) is designed, high classification accuracy is difficult to obtain for motor imagery (MI) electroencephalogram (EEG) signals in view of their relatively low signal-to-noise ratio. In this paper, a fused multidimensional classification method based on extreme tree feature selection (FMCM-ETFS) is proposed for discerning motor imagery EEG tasks. First, the EEG signal was filtered by a Butterworth filter for preprocessing. Second, C3, C4, and CZ channels were selected to extract time-frequency domain and spatial domain features using autoregressive (AR), common spatial pattern (CSP), and discrete wavelet transform (DWT). The extracted features were fused for a further feature elimination. Then, the features were selected using three feature selection methods: recursive feature elimination (RFE), principal component analysis method (PCA), and extreme trees (ET). The selected feature vectors were classified using support vector machines (SVM). Finally, a total of twelve subjects' EEG data from Inner Mongolia University of Technology (IMUT data), the 2nd BCI competition in 2003, and the 4th BCI competition in 2008 were employed to show the effectiveness of this proposed FMCM-ETFS method. The results show that the classification accuracy using the multidimensional fused feature extraction (AR + CSP + DWT) is 3%–20% higher than those using the aforementioned three single feature extractions (AR, CSP, and DWT). Extreme trees (ET), which is a sort of tree-based model method, outperforms RFE and PCA by 1%–9% in term of classification accuracies, when these three methods were applied to the procedure of feature extraction, respectively.

Multiattention Adaptation Network for Motor Imagery Recognition

Brain–computer interface (BCI) based on motor imagery electroencephalogram (EEG) has been widely used in various applications. Despite the previous efforts, the remained major challenges are effective feature extraction and the time-consuming calibration procedure. To address these issues, a novel multiattention adaptation network integrating the multiple attention mechanism and transfer learning is proposed to classify the EEG signals. First, the multiattention layer is introduced to automatically capture the dominant brain regions relevant to mental tasks without incorporating any prior knowledge about the physiology. Then, a multiattention convolutional neural network is employed to extract deep representation from raw EEG signals. Especially, a domain discriminator is applied to deep representation to reduce the differences between sessions for target subjects. The extensive experiments are conducted on three public EEG datasets (Dataset IIa and IIb of BCI Competition IV, and High Gamma dataset), achieving the competitive performance with average classification accuracy of 81.48%, 82.54%, and 93.97%, respectively. All the results outperform the state-of-the-art algorithms demonstrate the effectiveness and robustness of the proposed method. Importantly, we confirm that it is easier and more appropriate to transfer the information from local brain regions than from the whole brain. This enhances the transfer ability of deep features and, hence, it improves the performance of BCI systems.