Imagined character recognition through EEG signals using deep convolutional neural network

Abstract

Electroencephalography (EEG)-based brain computer interface (BCI) enables people to interact directly with computing devices through their brain signals. A BCI typically interprets EEG signals to reflect the user's intent or other mental activity. Motor imagery (MI) is a commonly used technique in BCIs where a user is asked to imagine moving certain part of the body such as a hand or a foot. By correctly interpreting the signal, one can perform a multitude of tasks such as controlling wheel chair, playing computer games, or even typing text. However, the use of motor-imagery-based BCIs outside the laboratory environment is limited due to the lack of their reliability. This work focuses on another kind of mental imagery, namely, the visual imagery (VI). VI is the manipulation of visual information that comes from memory. This work presents a deep convolutional neural network (DCNN)-based system for the recognition of visual/mental imagination of English alphabets so as to enable typing directly via brain signals. The DCNN learns to extract the spatial features hidden in the EEG signal. As opposed to many deep neural networks that use raw EEG signals for classification, this work transforms the raw signals into band powers using Morlet wavelet transformation. The proposed approach is evaluated on two publicly available benchmark MI-EEG datasets and a visual imagery dataset specifically collected for this work. The obtained results demonstrate that the proposed model performs better than the existing state-of-the-art methods for MI-EEG classification and yields an average accuracy of 99.45% on the two public MI-EEG datasets. The model also achieves an average recognition rate of 95.2% for the 26 English-language alphabets.

Electroencephalography based imagined alphabets classification using spatial and time‐domain features

“…They achieved the best classification accuracy by using CSP in different frequency bands by using an RF classifier. In recent work, Ullah and Halim 48 classified 26 imagined EEG signal alphabets using CNN. They calculated powers in various EEG frequency bands using Morlet wavelet transform and then used CNN for classification.…”

“…The imagined trials of all the subjects were mixed and then the accuracy was calculated. In some of the work in the literature, the subjects performed the same action for a longer duration of time 23–25,48 or performed a large amount of trials 10 which can lead to high accuracy. We have performed the classification from very limited data as the subjects were made to imagine a word only once in a trial.…”

“…The work in the literature for classifying all the imagined alphabets is very limited. Either the subject had to perform MI 24,25 which is limited in degrees of freedom and time consuming, or the BCI framework was subject dependent 48 . The existing work is also limited to the classification of vowels or few characters 6,7,10,11,23 .…”

A Quantized Convolutional Neural Network Implemented With Memristor for Image Denoising and Recognition

“…In the process of image perception, noise destruction will inevitably occur, which will seriously reduce the visual quality of the acquired image and further reduce the accuracy of image recognition ( Andrews and Hunt, 1977 ; Chatterjee and Milanfar, 2010 ). In addition to the traditional algorithm of designing filter for image denoising ( Ullah and Halim, 2021 ; Upadhyay et al, 2021 ), the current popular and effective algorithm is the deep learning algorithm for image denoising and feature extraction ( Zhang et al, 2017 ; Tian et al, 2020a , b ; Wang et al, 2020 ). However, the above algorithms all run on the von Neumann computer architecture; thus, a large amount of energy consumption will be required during the calculation process, while its computational efficiency is limited ( Chen et al, 2018 ; Xia and Yang, 2019 ; Zhang W. Q. et al, 2020 ).…”

Applying machine learning EEG signal classification to emotion‑related brain anticipatory activity

“…Emotions can be estimated from various physiological signals 17,18 , such as via skin conductance, electrocardiogram (ECG) and electroencephalogram (EEG). The latter has received a considerable amount of attention in the last decade, introducing several machine learning and signal processing techniques, originally developed in other contexts, such as text mining 19 , data processing 20 and brain computer interfaces 21,22 . Emotion recognition has been re-drawn as a machine learning problem, where proper EEG related features are used as inputs to specific classifiers.…”