Page 53 - IJEEE-2023-Vol19-ISSUE-1
P. 53

Al-Qaysi1, Al-Saegh, Hussein & Ahmed                                                                                             | 49

classification accuracy of 81.99% and 77.22% on BCIC IV        0.79% of classification accuracy on BCIC IV 2a and BCIC
2a and BCIC IV 2b datasets respectively.                       IV 2b respectively.

    Xu et. al. [23] proposed a method involving extracting         Kant et. al. [30] combined the continuous wavelet
features based on Hjorth parameter, the power spectrum         transform (CWT) with deep learning-based transfer learning
estimation, and time-frequency energy. Then sparse             and compared this method to different well-known deep
representation was used to acquire lower-dimensional           CNNs using the BCIC III 3a dataset. They obtained 95.71%
informative features while keeping the discriminative ability  as the best classification accuracy that is retained by the
among the different patterns. They obtained 79% of             VGG-19. Although there are several studies tried to improve
classification accuracy on BCIC IV 2b. Oh et. al. [24] used    the accuracy of classification, the results of the reviewed
Hjorth parameter and Fisher ratio to find the dominant         studies show that their still an area for enhancement.
frequency bands and the timing in training EEG signals and
79.1% of classification accuracy was achieved on BCIC IV                              III. METHODOLOGY
2b.
                                                                     The methodological framework of the WTNN for the
    Bagh and Reddy [25] used the Hilbert transform (HT) for    two-class MI EEG classification problem is presented in Fig.
the recognition of Event-related potentials, and the SVM for   1. This framework describes the whole process of pattern
decoding the MI signals. They obtained 86.11% and 82.50%       recognition starting from preparing the training samples and
of classification accuracy on BCIC III 3a and BCIC IV 2b       ending with the performance evaluation stage. The following
respectively.                                                  subsections give more details concerning the methodology
                                                               of this research.
    Zhu et. al. [26] The multi-channel input is encoded using
a separated channel convolutional network, and the encoded     A. MI EEG Datasets
features are concatenated and fed into a recognition network,
which performs the final MI task recognition. They obtained         A minimum number of channels is normally preferred
83% of classification accuracy on BCIC IV 2b.                  by developers in designing BCI-based systems such that they
                                                               can be easily employed with minimum cost for real-time
    Dai et. al. [27] proposed a segmentation data              applications [31]. Therefore, two MI EEG datasets recorded
augmentation method for MI EEG signals and used the new        by 3 channels are chosen in this study. The two datasets are
trails for training a CNN. They obtained 91.57% and 83% of     from the BCI competition datasets recorded at Graz
classification accuracy on BCIC IV 2a and BCIC IV 2b           University. More details regarding the two datasets are given
respectively.                                                  in the following subsections. The datasets consist of two
                                                               parts, namely the training part and the validation part. As
    Kim et. al. [28] proposed a method that rely on the power  such, it was used in developing and validating the WTNN to
spectral density (PSD) to find the noon-stationarity feature   deal with the complexity of the nine subjects’ specific brain
of each couple of EEG channels by calculating a matrix for     signals, such as inter and intra-subject differences. Given the
that. Such that, they exploited the time, frequency, and       lack of a large dataset to develop, evaluate, and validate a
spatial characteristics of the time-series signals. They       WTNN, the datasets of all the nine subjects were combined
obtained 89.36% of classification accuracy on BCIC IV 2a       (union) to form a large dataset for all the trials involving
for only two classes classification.                           different brain complexities.

    Sun et. al. [29] used an EOG channel for retaining the
potential related to MI tasks with the combination of Hjorth
algorithm to train their model. They obtained 76.45% and

Fig. 1: Methodological Framework for WTNN model
   48   49   50   51   52   53   54   55   56   57   58