Convolutional neural network with binary moth flame optimization for emotion detection in electroencephalogram

IAES International Journal of Artificial Intelligence (IJ-AI)
Vol. 13, No. 1, March 2024, pp. 1172~1178
ISSN: 2252-8938, DOI: 10.11591/ijai.v13.i1.pp1172-1178  1172
Journal homepage: http://ijai.iaescore.com
Convolutional neural network with binary moth flame
optimization for emotion detection in electroencephalogram
Tabarek Alwan Tuib1
, Baydaa Hadi Saoudi2
, Yaqdhan Mahmood Hussein1
, Thulfiqar H. Mandeel3
,
Fahad Taha Al-Dhief4
1
College of Engineering Electronics and Communication Engineering, Al Muthanna University, Samawah, Iraq
2
Department of ICT, Al-Samawa Technical Institute, Al-Furat Al-Awsat Technical University, Nafaq, Iraq
3
Department of Computer Techniques Engineering, College of Information Technology, Imam Ja’afar Al-Sadiq University, Baghdad,
Iraq
4
School of Electrical Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, UTM Johor Bahru, Johor, Malaysia
Article Info ABSTRACT
Article history:
Received Nov 26, 2022
Revised May 28, 2023
Accepted Jun 30, 2023
Electroencephalograph (EEG) signals have the ability of real-time reflecting
brain activities. Utilizing the EEG signal for analyzing human emotional
states is a common study. The EEG signals of the emotions aren’t distinctive
and it is different from one person to another as every one of them has
different emotional responses to same stimuli. Which is why, the signals of
the EEG are subject dependent and proven to be effective for the subject
dependent detection of the Emotions. For the purpose of achieving enhanced
accuracy and high true positive rate, the suggested system proposed a binary
moth flame optimization (BMFO) algorithm for the process of feature
selection and convolutional neural networks (CNNs) for classifications. In
this proposal, optimum features are chosen with the use of accuracy as
objective function. Ultimately, optimally chosen features are classified after
that with the use of a CNN for the purpose of discriminating different emotion
states.
Keywords:
Binary moth flame optimization
Classification
Convolutional neural networks
Electroencephalogram signals
Emotion detection
This is an open access article under the CC BY-SA license.
Corresponding Author:
Tabarek Alwan tuib
College of Engineering Electronics and Communication Engineering, Al Muthanna University
Samawah, Iraq
Email: tabarcka@gmail.com
1. INTRODUCTION
Brain-computer interfaces (BCI) are devices allowing the users to mentally manage a computer
program or Neuro prosthesis by translating brain activity into sequences of commands for the computer [1].
Human computer interaction became part of the daily life. In a similar way, the emotions are significant and
they constantly exist in the daily lives of the people. Emotions could provide numerous prospects in the
enhancement of interactions with the emotion-based computers, for example, the affective interactions for the
epileptic or autistic patients [2] It facilitates communication, particularly for physically challenged individuals
[3]. Emotions have an important impact on many tasks, including decision making, communication, and
human-computer interactions [4]. Emotion is a type of the unconscious or conscious feeling for a work or
phenomenon. Emotions are expressed by a variety of the physical and biological reactions, which include text,
voice, gestures, bio signals and facial expressions.
The people can’t hide their emotions although they do not want to show them. The reactions to the
emotions are a very significant key for communications amongst people [5] the detection of the emotions
represents one of the very important sub-areas of the affective computing, focusing upon the recognition of the
human emotions that are based upon various modalities, such body language, audio-visual expression,

Int J Artif Intell ISSN: 2252-8938 
Convolutional neural network with binary moth flame optimization for emotion … (Tabarek Alwan Tuib)
1173
physiological signals, and so on. In comparison with other Emotion detection is a significant affective
computing sub-area, focusing on the recognition of the human sentiments based upon various modalities, like
physiological signals, body language, audio-visual expressions, and so on. In comparison with other modalities,
physiological signals, like the EEG, electromyography (EMG), electrocardiogram (ECG), galvanic skin
response (GSR), and so on, have a benefit of being hard to disguise or conceal [6]. Compared to other outward
appearance indicators like the facial expression and gesture, the method based on EEG data is more dependable
among numerous approaches to emotion identification due to its objective evaluation and high accuracy [7].
There are several techniques for detecting brain activity, including magnetic resonance imaging (MRI) [8], [9]
magnetoencephalography (MEG) [10]. Off grid communication has been implemention in [11] and
electroencephalogram [12].
However, because the EEG has a quick reaction time and is less costly than other techniques, it is
commonly employed to monitor brain activity in brain-computer interfaces (BCI) research [12], [13]. By
inserting electrodes on the scalp, the EEG signals are captured as a weak potential and analyzed to create a BCI
system. The study has been based upon the recording and analyses of EEG brain activity, as well as the
recognition of EEG patterns linked with mental states [14]. Several strategies for designing a BCI system based
on EEG signals have been proposed, including event-related synchronizations [15] and event-related de-
synchronization [16]. A variety of EEG-based BCI systems were lately created in which feature extraction and
classification algorithms may identify EEG patterns in different mental states for information transmission [17].
Understanding brain's reactions to the variety of the emotional states could considerably develop the
computer models for the identification of the emotions. Numerous psychophysiology investigations [18]–[20]
showed relationships between human emotions and EEG patterns. In addition to that, with rapid development
of wearable devices and dry electrode approaches [21]–[24], there is now a possibility to transfer EEG-based
emotion recognitions from the laboratory to the real-world applications like the driving fatigue detections and
monitoring of the mental states [25]–[29].
Convolutional neural networks (CNNs) are a type of end-to-end model. End-to-end models that are
based upon the deep neural networks (DNNs) learn to map effectively from the main input to expected output
through the DNN. This prevents design and selections of complex manual features. However, utilizing the
CNNs for detecting the EEG emotions cannot directly accomplish the ideal results. The reason for this is that
the order of the input channels entering the CNNs must be meaningful. None-the-less, the main EEG channels
aren’t sorted according to their features. Channel proximity does not indicate the amount of channel
information. Which is why, the strategy of the increase of the amount of the information in the adjacent
channels through adjusting the channels one more time helps CNNs to learn effectively [30].
When the number of classified emotions is increased, the accuracy of EEG-based emotion detection
system is decreased. Since EEG contains noise and artifacts, emotion detection by EEG is still challenging. In
this paper, authors paid attention to emotion detection in EEG signals based of a binary moth flame
optimization (BMFO) algorithm for feature selection and CNN for classifications. In this proposed study,
optimum features have been selected with the use of the accuracy as objective function. Finally, optimally
chosen features will be classified after that with the use of the CNN for discriminating different emotion states.
2. PROPOSED METHOD
The emotion may be recognized mainly with a variety of the modalities such as the face images,
gestures and speech. Nonetheless, those methods of recognition have a susceptibility to person’s age, language,
culture, habit and appearance, which is why, they’re not universal and they lack the accuracy of the recognition.
In the present paper, the focus has been directed towards the recognition of emotions and connectivity analyses
with the CNNs. The suggested approach has 3 stages. In pre-processing stage, independent component analysis
(ICA) approach was used for removal of noise that is fundamentally induced electroencephalogram (EEG),
signals. For the purpose of achieving enhanced accuracy and high true positive rate the suggested system
introduced a binary moth-flame optimization (BMFO) [31] for feature selection and CNN for classification as
can see Figure 1. The emotion recognition performed by CNN with subject-independent connection features.
CNN can be defined as an End-to-End model type. End-to-end models that are based upon DNN effectively
learn mapping from original inputs to expected outputs via DNN. It avoids complicated manual feature design
and selections. However, directly utilizing the CNNs in the EEG emotion recognitions could barely accomplish
optimal results. It is because the channels’ order of inputs that are fed into the CNNs must be meaningful.
None-the-less, original EEG channels’ orders aren’t arranged based their characteristics. The channels’
proximity doesn’t reflect relevant information value between channels. Which is why, the strategy of the
increase of information amount on the adjacent channels through the rearrangements of the channels will help
the CNNs to learn in a more effective way. In the meantime, the features of the Pearson correlation coefficient
could be representing information of the connectivity between the variety of the channels of the EEG signal.

 ISSN: 2252-8938
Int J Artif Intell, Vol. 13, No. 1, March 2024: 1172-1178
1174
Figure 1. The flowchart of the classifications of the emotional states from the EEG data
2.1. Feature selection by BMFO algorithm
The moth flame optimization (MFO), which has been suggested by S. Mirjalili in 2015 [32], can be
defined as a new algorithm of optimization that has been inspired by moths’ special navigation approach in the
nature, which is referred to as the transverse orientation. The moths can be considered as fancy insects,
attracting towards an artificial light or the moonlight. Based on the MFO algorithm, the main MFO algorithm
moth modules (i.e., the actual search agents) and flames (i.e., optimal moth position). The distinct moth position
is updated based on the flame based on [31]. 𝑀𝑖 can given in (1):
𝑀𝑖 = 𝑆(𝑀𝑖, 𝐹𝑗) = 𝐷𝑖 ∗ 𝑒𝑏𝑡
∗ cos(2𝜋𝑡) + 𝐹𝑗 (1)
where Mi represents ith moth, S represents logarithmic spiral function, Fj denotes jth flame, Di represents
distance from Mi to Fj, b represents a constant value that defines logarithmic spiral shape and t represents some
random number in [−1, 1]. For the binary search space, moth positions are limited to binary variable. Which is
why, a sigmoidal function has been utilized for the purpose of transforming the position of every one of the
moths updated based on a flame (1) into the new position in the binary search space (2) for the BMFO.
𝑀𝑖 = {
1 𝑖𝑓 𝑟 𝑎𝑛𝑑 < 1
1+𝑒−(𝑋(𝐼+1))
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
(2)
Where X(I+1) represents real-value location update of moth for (I+1) th iteration and rand represents some
random number which is chosen in the (0, 1) interval.
3. EVALUATION
3.1. Data-set
The data-set that has been used in the experimentations is Shanghai Jiao Tong University (SJTU)
emotion data-base [33], containing 15 subjects’ EEG data (8 females + 7 males). In this data-set, EEG signals
have been recorded with the use of a 64 channel AgCl electrode at a 1,000 Hz sampling rate with the use of
international 10−20 system. Every electrode’s impendence has been < 5kΩ. SEED data-set was denoised by a
band pass filter and artefacts removal approach. And those denoised EEG signals have been divided to 5 brain
waves that correspond to frequency bands for all of the 62 channels. After that, brain waves have been separated
to 1 sec. fragments and DE features have been extracted. Based on the theory that higher bands of frequency
like β are more powerful and crucial in the detection of the EEG emotions. Figure 2 shows an example of
binaural and noise signals.
3.2. The initialization of parameters
For checking the quality of the suggested algorithm, we adjusted CNN parameters. We did this task
by considering 30 for maximum iteration, training percentage was considered for 80% of the data, and test
percentage was taken into account for 20% of the data. Table 1 shows the settings of the parameters (CNN).
3.3. The evaluation criteria
The choice of criterion for evaluating the efficiency of the approach is dependent upon the problem
that needs being solved. We assume that some data samples are available. These data are given separately to

1175
the model and a class is received as output for every one of them. The predicted class of the model and the
actual class of data have been listed in a Table 2 that has been referred to as the confusion matrix.
Figure 2. Plot of binaural and noise signals
Table 1. Initial values for parameters on CNN
Parameter Initial parameters
Maximum iteration
Training percentage
Test percentage
Input layers
Learning rate
30
80%
20%
[64,128,1]
0.001
Table 2. Confusion matrix
N
P
𝑻𝑷
𝑷𝑷
precision
FP
False positives
TP
True positives
PP
Predicted positives
TN
True negatives
FN
False negatives
PN
Negative predictions
𝑻𝑵
𝑵
feature
𝑻𝑷
𝑷
Recall/sensitivity
Section, it provides an exact description and definition of the confusion matrix. From this table, four
simple criteria can be obtained directly: TP (i.e., true positive) and TN (i.e., true negative) denote the number
of the positive and negative samples which have been accurately classified, while FP (false positive) and FN
(false negative) are the number of the positive and negative samples which have been incorrectly classified,
− True positive rate: TPR (true positive rate) = TP/(TP+FN). Rate of the positive samples that are
correctly classified in positive class. It is also called a recall or sensitivity.
− True negative rate: TNR (correct negative rate) = TN/(FP+TN). Rate of the negative samples that are
correctly classified in the negative class are also called feature.
− False Positive Rate: FPR (false positive rate) = FP/(FP+TN). Rate of the negative samples that are
incorrectly classified in the negative class.
− False negative rate: FNR (false negative rate) = FN/(TP+FN). Percentage of the positive samples that
are incorrectly classified in the negative class.

 ISSN: 2252-8938
1176
3.3.1. Accuracy criterion
The most common criterion for classifying accuracy and vice versa is the error rate. This one is the
ratio of positive samples that are actually positive and shows the accuracy of the learning model.
Mathematically, this ratio can be described in (3),
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =
𝑇𝑃+𝑇𝑁
𝑇𝑃+𝐹𝑃+𝐹𝑁+𝑇𝑁
(3)
3.4. The evaluation of results
Emotions are characterized by the EEG signal with a variety of the feature extraction approaches and
classification algorithms. The accuracy of the emotions could vary from one extract method to the other. After
the implementation of the proposed approach in MATLAB 2018b environment and using the sama plot
program, the results of both accuracy and detection rate were plotted, and the sorted results from the proposed
improved neural network are shown in Table 3. CNN, have been compared. We compared the accuracy of
current and proposed approaches for discriminatory emotion state. Compared with the existing methods, the
error rate is reduced and the accuracy is superior to the existing approaches. In addition, the proposed network
structure is simple and the memory consumption parameters are few. We also confirm that the shallow network
also has a relatively good detection effect. In Figures 3 and 4, the trend of increasing accuracy and decreasing
losses over 50 epochs.
Table 3. The comparison accuracy results of the suggested approach with the level of various techniques
S .no Feature Selection Classify method Accuracy
1 BMFO CNN 99.4%
2 Firefly Algorithm ISO-FLANN Classification 95 %
3 DWT K-NN 83.26%
4 Dynamic features and PCA SVMs 64.70 _ 82.91%
5 WT SVMs 82.38 %
6 HOC SVMs 82.33%
7 STFT SVMs 80%
8 HOC NN 80.50%
9 Connectivity Features And PCA RBF 84.60%
10 DWT LDA 75.21%
11 HOC QDY 62.3 %
Figure 3. The accuracy results of the proposed
approach during 50 epochs
Figure 4. The loss results of the proposed approach
during 50 epochs
4. CONCLUSION
We introduced emotion recognition based on EEG algorithm. In this algorithm, emotion is associated
with the EEG signals. were the proposed system introduced a binary moth flame optimization (BMFO)
algorithm for feature selection and convolutional neural networks (CNN) for classification. In the suggested
study, optimum features had been chosen with the use of the accuracy as objective function. Finally, optimally
chosen features will be classified after that with the use of a CNN for discriminating a variety of the states of
emotions. We compared the accuracy of current and suggested approaches for discriminatory emotion state.
Compared with the existing methods, the error rate is reduced and the accuracy is better than the existing
approaches. In addition, the suggested network structure is simple and the memory consumption parameters
are few.

1177
REFERENCES
[1] S. Nasehi and H. Pourghassem, “A new feature dimensionally reduction approach based on general tensor discriminant analysis in
EEG signal classification,” in Proceedings - 2011 International Conference on Intelligent Computation and Bio-Medical
Instrumentation, ICBMI 2011, Dec. 2011, pp. 188–191. doi: 10.1109/ICBMI.2011.32.
[2] R. M. Mehmood and H. J. Lee, “EEG based emotion recognition from human brain using Hjorth parameters and SVM,”
International Journal of Bio-Science and Bio-Technology, vol. 7, no. 3, pp. 23–32, Jun. 2015, doi: 10.14257/ijbsbt.2015.7.3.03.
[3] A. Walendziak, “Nontrivial BCK/BCI-algebras do not satisfy the fuzzy ascending chain condition,” Fuzzy Sets and Systems, vol.
158, no. 8, pp. 922–923, Apr. 2007, doi: 10.1016/j.fss.2006.11.018.
[4] M. Murugappan, N. Ramachandran, and Y. Sazali, “Classification of human emotion from EEG using discrete wavelet transform,”
Journal of Biomedical Science and Engineering, vol. 03, no. 04, pp. 390–396, 2010, doi: 10.4236/jbise.2010.34054.
[5] H. Yang, J. Han, and K. Min, “A multi-column CNN model for emotion recognition from EEG signals,” Sensors (Switzerland),
vol. 19, no. 21, p. 4736, Oct. 2019, doi: 10.3390/s19214736.
[6] P. Zhong, D. Wang, and C. Miao, “EEG-based emotion recognition using regularized graph neural networks,” IEEE Transactions
on Affective Computing, vol. 13, no. 3, pp. 1290–1301, Jul. 2022, doi: 10.1109/TAFFC.2020.2994159.
[7] F. T. Al-Dhief et al., “Voice Pathology detection using machine learning technique,” in 2020 IEEE 5th International Symposium
on Telecommunication Technologies, ISTT 2020 - Proceedings, Nov. 2020, pp. 99–104. doi: 10.1109/ISTT50966.2020.9279346.
[8] I. Prohovnik, P. Skudlarski, R. K. Fulbright, J. C. Gore, and B. E. Wexler, “Functional MRI changes before and after onset of
reported emotions,” Psychiatry Research - Neuroimaging, vol. 132, no. 3, pp. 239–250, Dec. 2004, doi:
10.1016/j.pscychresns.2004.03.005.
[9] J. Zhang, G. Sudre, X. Li, W. Wang, D. J. Weber, and A. Bagic, “Clustering linear discriminant analysis for MEG-based brain
computer interfaces,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 19, no. 3, pp. 221–231, Jun. 2011,
doi: 10.1109/TNSRE.2011.2116125.
[10] A. S. Garrett and R. J. Maddock, “Separating subjective emotion from the perception of emotion-inducing stimuli: An fMRI study,”
NeuroImage, vol. 33, no. 1, pp. 263–274, Oct. 2006, doi: 10.1016/j.neuroimage.2006.05.024.
[11] Y. M. Hussain et al., “Smartphone’s off grid communication network by using Arduino microcontroller and microstrip antenna,”
Telkomnika (Telecommunication Computing Electronics and Control), vol. 19, no. 4, pp. 1100–1106, Aug. 2021, doi:
10.12928/TELKOMNIKA.v19i4.15949.
[12] C. Guger et al., “How many people are able to control a P300-based brain-computer interface (BCI)?,” Neuroscience Letters, vol.
462, no. 1, pp. 94–98, Sep. 2009, doi: 10.1016/j.neulet.2009.06.045.
[13] S. Nasehi and H. Pourghassem, “Online mental task classification based on DWT-PCA features and probabilistic neural network,”
International Journal of Imaging and Robotics, vol. 7, no. 1, pp. 110–118, 2012.
[14] H. J. Hwang, K. Kwon, and C. H. Im, “Neurofeedback-based motor imagery training for brain-computer interface (BCI),” Journal
of Neuroscience Methods, vol. 179, no. 1, pp. 150–156, Apr. 2009, doi: 10.1016/j.jneumeth.2009.01.015.
[15] L. Leocani et al., “Movement preparation is affected by tissue damage in multiple sclerosis: Evidence from EEG event-related
desynchronization,” Clinical Neurophysiology, vol. 116, no. 7, pp. 1515–1519, Jul. 2005, doi: 10.1016/j.clinph.2005.02.026.
[16] L. I. Aftanas, A. A. Varlamov, N. V. Reva, and S. V. Pavlov, “Disruption of early event-related theta synchronization of human
EEG in alexithymics viewing affective pictures,” Neuroscience Letters, vol. 340, no. 1, pp. 57–60, Apr. 2003, doi: 10.1016/S0304-
3940(03)00070-3.
[17] D. Coyle, T. M. McGinnity, and G. Prasad, “Improving the separability of multiple EEG features for a BCI by neural-time-series-
prediction-preprocessing,” Biomedical Signal Processing and Control, vol. 5, no. 3, pp. 196–204, Jul. 2010, doi:
10.1016/j.bspc.2010.03.004.
[18] D. Sammler, M. Grigutsch, T. Fritz, and S. Koelsch, “Music and emotion: Electrophysiological correlates of the processing of
pleasant and unpleasant music,” Psychophysiology, vol. 44, no. 2, pp. 293–304, Mar. 2007, doi: 10.1111/j.1469-8986.2007.00497.x.
[19] G. G. Knyazev, J. Y. Slobodskoj-Plusnin, and A. V. Bocharov, “Gender differences in implicit and explicit processing of emotional
facial expressions as revealed by event related theta synchronization,” Emotion, vol. 10, no. 5, pp. 678–687, Oct. 2010, doi:
10.1037/a0019175.
[20] D. Mathersul, L. M. Williams, P. J. Hopkinson, and A. H. Kemp, “Investigating models of affect: Relationships among EEG alpha
asymmetry, depression, and anxiety,” Emotion, vol. 8, no. 4, pp. 560–572, 2008, doi: 10.1037/a0012811.
[21] C. Grozea, C. D. Voinescu, and S. Fazli, “Bristle-sensors - Low-cost flexible passive dry EEG electrodes for neurofeedback and
BCI applications,” Journal of Neural Engineering, vol. 8, no. 2, p. 25008, Mar. 2011, doi: 10.1088/1741-2560/8/2/025008.
[22] Y. M. Chi, Y. Te Wang, Y. Wang, C. Maier, T. P. Jung, and G. Cauwenberghs, “Dry and noncontact EEG sensors for mobile brain-
computer interfaces,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 20, no. 2, pp. 228–235, Mar.
2012, doi: 10.1109/TNSRE.2011.2174652.
[23] L. F. Wang, J. Q. Liu, B. Yang, and C. S. Yang, “PDMS-based low cost flexible dry electrode for long-term EEG measurement,”
IEEE Sensors Journal, vol. 12, no. 9, pp. 2898–2904, Sep. 2012, doi: 10.1109/JSEN.2012.2204339.
[24] Y. J. Huang, C. Y. Wu, A. M. K. Wong, and B. S. Lin, “Novel active comb-shaped dry electrode for eeg measurement in Hairy
site,” IEEE Transactions on Biomedical Engineering, vol. 62, no. 1, pp. 256–263, Jan. 2015, doi: 10.1109/TBME.2014.2347318.
[25] M. Maqsood, S. A. Lashari, M. A. Saare, S. A. Sari, Y. M. Hussein, and H. O. Hatem, “Minimization Response Time Task
scheduling Algorithm,” IOP Conference Series: Materials Science and Engineering, vol. 705, no. 1, p. 12008, Nov. 2019, doi:
10.1088/1757-899X/705/1/012008.
[26] N. H. Liu, C. Y. Chiang, and H. M. Hsu, “Improving driver alertness through music selection using a mobile EEG to detect
brainwaves,” Sensors (Switzerland), vol. 13, no. 7, pp. 8199–8221, 2013, doi: 10.3390/s130708199.
[27] J. B. F. Van Erp, F. Lotte, and M. Tangermann, “Brain-computer interfaces: Beyond medical applications,” Computer, vol. 45, no.
4, pp. 26–34, Apr. 2012, doi: 10.1109/MC.2012.107.
[28] B. J. Lance, S. E. Kerick, A. J. Ries, K. S. Oie, and K. McDowell, “Brain-computer interface technologies in the coming decades,”
Proceedings of the IEEE, vol. 100, no. SPL CONTENT, pp. 1585–1599, May 2012, doi: 10.1109/JPROC.2012.2184830.
[29] P. Aspinall, P. Mavros, R. Coyne, and J. Roe, “The urban brain: Analysing outdoor physical activity with mobile EEG,” British
Journal of Sports Medicine, vol. 49, no. 4, pp. 272–276, Mar. 2015, doi: 10.1136/bjsports-2012-091877.
[30] Z. Wen, R. Xu, and J. Du, “A novel convolutional neural networks for emotion recognition based on EEG signal,” in 2017
International Conference on Security, Pattern Analysis, and Cybernetics, SPAC 2017, Dec. 2018, vol. 2018-January, pp. 672–677.
doi: 10.1109/SPAC.2017.8304360.
[31] L. Kumar and K. K. Bharti, “An improved BPSO algorithm for feature selection,” in Lecture Notes in Electrical Engineering, vol.
524, Springer Singapore, 2019, pp. 505–513. doi: 10.1007/978-981-13-2685-1_48.
[32] S. Mirjalili, “Moth-flame optimization algorithm: A novel nature-inspired heuristic paradigm,” Knowledge-Based Systems, vol. 89,

 ISSN: 2252-8938
1178
pp. 228–249, Nov. 2015, doi: 10.1016/j.knosys.2015.07.006.
[33] W. L. Zheng and B. L. Lu, “Investigating critical frequency bands and channels for EEG-based emotion recognition with deep
neural networks,” IEEE Transactions on Autonomous Mental Development, vol. 7, no. 3, pp. 162–175, Sep. 2015, doi:
10.1109/TAMD.2015.2431497.
BIOGRAPHIES OF AUTHORS
Tabarek Alwan Tuaib was born in sammawah, Iraq, in 1996. He received a
B.S in computer techniques/computer electronics engineering in 2017-2018 from the
Electrical Engineering Technical College and an M.S. degree in Computer Engineering,
Artificial Intelligence, and Robotics from Islamic Azad University, in 2021, respectively. She
studies a Ph.D. degree in Computer Engineering, Artificial Intelligence in University of
Tabriz. Her current research interests include deep learning, machine learning, artificial
intelligence. She can be contacted at email: tabarcka@gmail.com.
Baydaa Hadi Saoudi was born in sammawah, Iraq, in 1970. She received a B.S.
in Computer Science in 1990-1991 from Technology University Baghdad City. I am studying
for a master's degree in Artificial Intelligence in Iran from Isfahan Azad Universityt. She can
be contacted at email: ins.byd44@atu.edu.iq.
Yaqdhan Mahmood Hussein was born in sammawah, Iraq, in 1991. He
received the B.S in computer techniques engineering in 2014-2015 from Islamic University
College in Najaf city. and M.S. degrees in Electronic Engineering (Telecommunication
System) from University Technical Malaysia Melaka, Malaysia, in 2018, respectively. He
studies the Ph.D. degree in Electronic Engineering in University Technology Malaysia in
Johor Bahru city, His current research interests include millimeter wave antennas, base
station antennas, and SIW technology with butler matrix and beamforming. He can be
contacted at email: Yaqthanm79@gmail.com.
Thulfiqar H. Mandeel was born in Iraq in 1991. He received the M.Sc. degree
in Embedded System Design Engineering and Ph.D. degree in Computer Engineering from
Universiti Malaysia Perlis (UniMAP) in Perlis, Malaysia in 2015 and 2019 respectively. He
joined UniMAP in 2016 as a graduate assistant, and is currently a lecturer in Imam Ja’afar
Al Sadiq University in Iraq. His current research interests include image processing,
biometric recognition, medical image classification, and deep learning. He can be contacted
at email: Thulfiqar.hussein@sadiq.edu.iq.
Fahad Taha Al-Dhief was born in Iraq, Amarah city. He received the B.S. in
Software Engineering from Imam Jaafar Al-Sadiq University, Iraq in 2013 and also received
M.S. in Computer Science from University Kebangsaan Malaysia, Malaysia in 2016.
Currently, he is a Ph.D. student at Universiti Teknologi Malaysia, Faculty of Electrical
Engineering, Department of Communication Engineering, Malaysia. He is an active student
member of IEEE, and a member of IEEE Communications Society. His research interests are
sensor network, routing protocols, mobile ad-hoc network, social network, internet of things,
machine learning, artificial neural networks, deep learning, and location-based service. He
can be contacted at email: Taha-1989@graduat.utm.my.

Convolutional neural network with binary moth flame optimization for emotion detection in electroencephalogram

More Related Content

Similar to Convolutional neural network with binary moth flame optimization for emotion detection in electroencephalogram (20)

More from IAESIJAI (20)

Recently uploaded (20)

Convolutional neural network with binary moth flame optimization for emotion detection in electroencephalogram