Guest Editorial: Current Trends in Cognitive Science and Brain Computing Research and Applications

Varun Bajaj, G R Sinha, Siuly Siuly, Abdulkadir Şengur
2020 Electronics Letters  
This is our first Special Issue in Electronics Letters on Current Trends in Cognitive Science and Brain Computing Research and Applications. This issue carries 18 Letters from multinational contributors highlighting their novel findings, substantial analysis and critical reviews on the topic. With the rapid advancement of technology, studies on brain signal analysis are facing increasing challenges to acquire, analyse, and apply the large amount of knowledge necessary to solve complex problems
more » ... n the fields of health and medicine. As a result of the ongoing COVID-19 pandemic, millions of people are experiencing mental stress and psychological changes, and for this reason research on the cognitive assessment of various mental activities in the human brain becomes ever more relevant. Cognitive science deals with brain signal research and its interpretation, enabling its application in all philosophical and psychological aspects of the human brain, while cognitive science and brain computing deals with the interpretation of brain signals and decision making based on the signal description, with various imaging modalities, such as functional magnetic resonance imaging (fMRI) and electroencephalogram (EEG), being used for the interpretation of brain status. Current trends in these areas of research focus on the assessment of cognitive ability, and there has been limited but significant research exploring how the replication of human brain function can be applied in the fields of AI, driverless vehicles, robotics, healthcare, and hazardous site applications. The main objectives of this Special Issue which are now achieved as outcomes are: study the impact of cognitive science and brain computing based research using modern machine learning tools; investigate the research in the area of the cognitive ability of the human brain using biomedical signal analysis; and study of application of brain computer interface (BCI) research. Based on critical review, we have accepted 18 Letters in this Special Issue from the researchers who have investigated various aspects of cognitive science and brain computing, which includes processing of EEG signal and its use in emotion classification, epileptic seizure analysis, study of drowsiness etc., and MRI related issues. Brief descriptions of the contributions of these chapters are as follows. Khare et al. present classification of emotions using time-order representation which is further based on the S-transform and convolutional neural network (CNN). The identification of human emotions is investigated for improvement of BCI. This work includes various types of emotions such as sadness, fear, happiness etc. and the proposed method has been proved outperforming as compared to existing work tested on similar datasets. In 'Efficient approach for EEG-based emotion recognition', Şengür and Siuly also investigate emotion recognition utilising deep features and continuous wavelet transform (CWT). The extracted rhythm signals are converted into the EEG rhythm images by employing the CWT. Deep features using long short term memory (LSTM) method are employed for classification of emotions mainly into two categories "Arousal" and "Valence". Alakus and Turkoglu, in 'Emotion recognition with deep learning using GAMEEMO data set', employ BiLSTM and apply it over GAMEEMO dataset; and the spectral entropy values of the EEG signals of all channels are calculated. These values are classified by using BiLSTM architecture to predict positive and negative emotions. In 'Motor imagery BCI classification based on novel two-dimensional modelling in empirical wavelet transform', Sadiq et al. implement motor imagery BCI classification using two dimensional modelling in empirical wavelet transform. The sum of distance from each point relative to a coordinate center (SDTC) is extracted from 2D modelling of modes, which is used as input to feedforward neural network and cascade forward neural networks for classification of MI tasks. In 'Classification of epileptic EEG signals using sparse spectrum based empirical wavelet transform', Nishad et al. propose sparse spectrum based empirical wavelet transform (SS-EWT) that aims to decompose the EEG signal into coefficients. The SS-EWT coefficients, the cross information potential (CIP) and normalised energy (NE) are extracted as features. After ranking, these features are fed into the k-nearest neighbour (k-NN) classifier to classify EEG signals corresponding to different brain activities. In 'Diagnosis of autism spectrum disorder from EEG using a time-frequency spectrogram image-based approach', Tawhid et al. suggest diagnosis of autism spectrum disorder by pre-processing EEG signals which are converted to two-dimensional (2D) images using short-time fourier transform (STFT). The textural features are extracted, and significant features are selected using principle component analysis (PCA), and fed to support vector machine (SVM) classifier for classification. Singh and Singh have contributed an article entitled 'Realising transfer learning through convolutional neural network and support vector machine for mental task classification'. This article proposes a framework based on transfer learning for mental task classification where pre-trained network is used for the extraction of diverse features and SVM is employed for classification. In another interesting article, 'A novel approach based on wavelet packet transform and 1d-RMLBP for drowsiness detection using EEG', Alçin reports an EEG based approach for detection of drowsiness. The proposed method consists of random sampling-based artificial signal augmentation, wavelet packet transform (WPT) decomposition, logarithmic energy entropy (LEE), and one-dimension region mean local binary pattern (1d-RMLBP) based features and the features are tested by kNN and SVM classifier. The article by Ari entitled 'Analysis of EEG signal for seizure detection based on WPT' suggests a design to create a computer-based expert system for the detection of epilepsy. In this Letter, WPT is employed for calculation of approximation and detail coefficients of EEG signals. The coefficients are subjected to feature extraction using dispersion entropy (DE) and line length (LL) methods, and the extracted feature vectors are used as input to SVM and kNN classifiers. The article entitled 'Automatic drowsiness detection using electroencephalogram signal' by Dutta et al. presents an impressive idea based on clustering variational mode decomposition (CVMD) for automatic detection of drowsiness from EEG signals. The oscillatory mode characteristics of the clusters are extracted in terms of several features. These features are fed as input into the least-squares support vector machine (LS-SVM) classifier. Mandal et al., in the Letter 'Classification of working memory loads using hybrid EEG and fNIRS in machine learning paradigm', explore the different classification working memory load levels using a hybrid BCI system. N-back cognitive tasks like 0-back, 2-back, and 3-back are used to create working memory load on participants while recording EEG and functional near-infrared spectroscopy (fNIRS) signals simultaneously. A combination of statistically significant features obtained from EEG and fNIRS corresponding to frontal region channels are used to classify different N-back commands by SVM classifier. The Letter entitled 'Protection of BCI system via reversible watermarking of EEG signal' by Bhalerao et al. introduces a security block in the BCI system that ensures that the recorded EEG signals are intact at the receiving end of wireless BCI system. The proposed scheme works efficiently and performance of BCI system remains unaltered after inclusion of security block. In their Letter 'Brain-computer interface-based single trial P300 detection for home environment application', Shukla et al. develop a P300 speller-based BCI which helps disabled people ease their lives by accessing mobile, light, fan, door, television, electric heater, etc. The authors propose a single trial weighted ensemble of compact convolution neural network (WE-CCNN), and obtain an ITR of 46.45 bits per minute and an average target appliance accuracy of 93.22% for BCI-based home environment system. In 'Brain MRI-based Wilson disease tissue classification: an optimised deep transfer learning approach', Saba et al. present a computer-aided design CADx-based automated classification strategy that uses optimised transfer learning (TL) utilising two novel paradigms known as (i) MobileNet, and (ii) the Visual Geometric . Authors use four-fold augmentation and report that the VGG-19 is superior to MobileNet demonstrating accuracy and AUC pairs as: 95.46+7.70%, 0.932 (p<0.0001) and 86.87+2.23%, 0.871 (p<0.0001), respectively. In 'Robust spatial information based tumour detection for brain MR images', Maharana et al. propose a novel unsupervised spatial information
doi:10.1049/el.2020.2790 fatcat:tyub3uxoyngivddcg7f6esce4y