An efficient optimization technique of EEG

An efficient optimization technique of EEG decomposition for user authentication system Zaid Abdi Alkareem Alyasseri∗† , Ahamad Tajudin Khader∗ , Mohammed Azmi Al-Betar‡ , Joao P. Papa§ , Osama Ahmad Alomari∗ , Sharif Naser Makhadme∗ ∗ School

of Computer Sciences, Universiti Sains Malaysia, Pulau Pinang, Malaysia Department-Faculty of Engineering, University of Kufa, Najaf, Iraq ‡ Depart. of IT, Al-Huson University College, Al-Balqa Applied University, Al-Huson, Irbid, Jordan § San Paulo State University, Department of Computing, Bauru, Brazil † ECE

Abstract—Since past three decades, the world is transformed into digital society, where every individual is living with a unique digital identifier. The main purpose of this identifier is to distinguish from others as well as to deal with digital machines which are surrounding the world. Recently, many researchers proved that the brain electrical activity or electroencephalogram (EEG) signals is able to provide a robust and unique features which can be considered as a new biometric authentication technique. One of the most important things to extract the efficient unique features from the input EEG signals is to find the optimal method to decompose the input EEG signals. Therefore, this paper proposed a novel method for EEG signal denoising based on multi-objective flower pollination algorithm with wavelet transform (MOFPA-WT) to extract the efficient features from denoised signals. MOFPA-WT is tested using a standard EEG signal dataset, namely, Keirn EEG datasetwhich has five mental tasks, includes baseline, multiplication two numbers, geometric figure rotation, letter composing, and visual counting. The performance of MOFPA-WT is evaluated using three criteria, namely, accuracy, true acceptance rate (TAR), and false acceptance rate (FAR). It is worth mentioning that the proposed method achieves the highest accuracy result which can be obtained using mental tasks based on geometric figure rotation compared with mental tasks. Index Terms—EEG, Biometric, Authentication, Flower pollination algorithm, multi-objective

I. I NTRODUCTION Electroencephalogram (EEG) is a graphical recording of brain electrical activity that is recorded from the scalp. This recording represents the voltage fluctuations resulting from ionic current flows within the neurons of the brain [15], [23]. Therefore, EEG signals can provide most of the required information about brain activity. EEG signals from the brain are captured using invasive or non-invasive techniques [18]. The main difference between these techniques is that the invasive approach involves the use of electrode arrays implanted inside the brain, such as ECoG BCI for arm movement control [19]. Hans Berger in [12] proposed the first use of EEG signals as a non-invasive technique for capturing brain activities. Over the past several decades, researchers have developed Hans’s technique to suit multiple applications. For example, EEG Corresponding author: Zaid [email protected]

Abdi

Alkareem

Alyasseri

(email:

signals have been used in medical applications for prevention, detection diagnosis, rehabilitation and restoration of patients. The EEG has also been used for non-medical applications, such as education and self-regulation, neuromarketing and advertisement, neuroergonomics and smart environment, games and entertainment, and learning and education as summarized in [1]. Recently, EEG signals have been successfully used as a new biometric technique in security and authentication applications [1], [15]. The researchers found the EEG signals are providing a robust and unique features which can be considered as a new biometric technique as well as it can not be spoofed because until now there is no known method for attacking the EEG signal [18]. Pinki Kumar et. al. in [15] proposed a user identification system based on EEG signal collected from six users using EMOTIVE EPOC headset which has 14 channels. In the preprocessing phase, they used a Butterworth 5th order filter with range 6-35 Hz to achieve the highest signal-to-noise ratio (SNR) of the input EEG signal. In feature extraction phase, wavelet transform (WT) technique is proposed to extract the unique features of EEG signal. In addition, three basic statistical measurements extracted from EEG signal includes Mean, Standard deviation, and Energy for each sub-bands rhythms (i.e., high gamma, gamma, alpha, beta, theta, and delta) as unique features to pass these features to next phase. For classification phase, LVQ-NN classifier used to recognize the users. Finally, the recognition rate has been calculated over the different scenarios to find the best combination of channels which can provide the correct classification. Later, same authors proposed a novel method for EEG features generation [16]. The proposed applied a canonical correlation analysis (CCA) method for features fusion level to improve the recognition rate of the identification technique. The proposed method tested using a standard EEG dataset [14] which has five mental tasks, includes baseline, multiplication two numbers, geometric figure rotation, letter composing, and visual counting each task repeated several times for ten seconds and the EEG signals collected from seven subjects. The proposed method used three techniques to extract the features from the input EEG signals which are empirical mode decomposition, information theoretic measure, and statistical

measurement. In order to classify the extracted features linear vector quantization (LVQ) neural network classifier and its extension (LVQ2) has been used to find the accuracy rate based on different mental tasks. The performance of the proposed method provided better results compared with a basic method for features fusion. Rodrigues et. al. in [20] used binary flower pollination algorithm [24] to obtain the best channel which can provide the highest recognition rate for person identification based on EEG signal. Their work was tested using a standard EEG datatsets which are motor / movement and imaginary [22]. Finally, their work able to obtain the highest recognition rate equal (87%) with reducing the number of EEG channels to half. The main objective of this paper is to propose a multiobjective flower pollination algorithm with wavelet transform (MOFPA-WT) to decompose the input EEG signal to find the optimal features which can achieve the highest accuracy. MOFPA-WT applied using two objective functions which are: min(MSE) and max(SNR) to obtain the best combination of WT parameters for EEG signal denoising. The proposed method is implemented according to the weighted sum approach to combine multi-objectives into a composite one objective function. The original EEG signal is taken from a standard EEG dataset which is Keirn EEG dataset1 which has five mental tasks, includes baseline, multiplication two numbers, geometric figure rotation, letter composing, and visual counting each task repeated several times for ten seconds and the EEG signals collected from seven subjects [14]. The original EEG signals decompose into five level to extract the unique features from each sub-bands (i.e. high gamma, gamma, alpha, beta, theta, and delta) where four features are extracted which are namely: Mean, Standard deviation, Entropy, and Energy. For evaluating the performance of MOFPA-WT, the results are evaluated in terms of three measurement factors: accuracy, true acceptance rate (TAR), and false acceptance rate (FAR). It is worth mentioning that the proposed method achieves the highest accuracy result which can be obtained using mental tasks based on geometric figure rotation compared with mental tasks. This paper is organized as follows. Section II describes a wavelet transform and the principal of WT for EEG signal denoising. Section III provides a background about the flower pollination algorithm and its multi-objective technique. Section IV describes the proposed system. The results and discussion describes in section V. Finally, the conclusion and future works describes in section VI. II. EEG S IGNAL DENOISING USING WAVELET T RANSFORM Wavelet Transform (WT) is a powerful and common tool for time-frequency domain signal representation. WT has successfully applied for signal compression, feature extraction and selection, and others [5]–[8]. In general, the WT can be classified into two types: discrete wavelet transform (DWT) 1 http://www.cs.colostate.edu/eeg/main/data/1989

Keirn and Aunon

and continuous wavelet transform (CWT) [8]. In the recent few years, the WT has been extensively used with non-stationary signals, such as ECG and EEG because the WT shown a powerful outcomes in removing several EEG artifact noises and extracting the EEG features [8]. In this paper, the DWT has been used to decompose the input EEG signal to extract unique features from each EEG sub-bands (i.e., high gamma, gamma, alpha, beta, theta, and delta). One of the popular methods for DWT is proposed in [13] and so-called Donoho’s approach which extracted as follows: X C(a, b) = x(n)gj,k (n) (1) n∈Z

where C(a,b) denotes the wavelet dynamic coefficients, a = 2−j , b = k2−j , j ∈ Z, k ∈ Z; a is the size of the time scale, b is the translation, x(n) is the input EEG signal, and gj,k (n) = 2j/2 g(2j n − k) is the DWT. The task of DWT is to decompose the input signal using different coefficients levels to correct the high frequency of the input signal. The denoising process involves three phases: • EEG signal decomposition, the original EEG signal will be divided into five levels, at each level the EEG signal will be decomposed into two parts namely Approximation coefficients (cA), and Detail coefficients (cD). The cD will process using high-pass filter and cA will be continuously decomposed for next level. • Thresholding where for each level a threshold value defined according to the coefficients noise level. • Reconstruction, the EEG denoised signal is reconstructed using inverse discrete wavelet transform iDWT. The WT has five parameters where each parameter has different types (See Table I). The efficiency of noise reduction and unique features extraction relies on the selection of wavelet parameters. The wavelet denoising process has three phases: The first phase is the decomposition of the EEG signal using DWT. This phase involves selecting the appropriate mother wavelet function (Φ) for use in the EEG signal decomposition task. The second wavelet parameter, that is, the decomposition level (L), is also selected in this phase normally based on the EEG signal and experience. It should be noted that the selection of appreciate parameters of WT (which is one of the main goal of this paper) is recently accomplished using optimization techniques such FPA, β-hill climbing (βhc), and genetic algorithm (GA) [6], [8]. In the second phase, thresholding is applied. The wavelet provides two standard types of thresholding functions (β), namely, hard and soft thresholding [13]. The thresholding type (soft or hard), selection rules (λ), and rescaling methods (ρ) must all be selected. These threshold mechanisms must be applied because the selection will affect the global denoising performance. The thresholding value is generally defined based on the standard deviation (σ) of the noise amplitude. The wavelet parameters (β, λ, and ρ) must be separately applied for each wavelet coefficient (cA and cD) level. In the last phase, the denoised EEG signal is reconstructed by iDWT.

TABLE I: Ranges of the WT denoising parameters Wavelet denoising parameters Wavelet function Φ Thresholding function β Decomposition level L Thresholding selection rule λ Rescaling approach ρ

IV. P ROPOSED S YSTEM

Method (range) Symlet (sym1..sym45), Coiflet (coif1..coif5), Daubechies (db1..db45), and Biorthogonal (bior1.1.. bior1.5&bior2.2 .. bior2.8& bior3.1..bior3.9). soft or hard 5 Heursure, Rigsure, Sqtwolog, and Minimax one, sln, and mln

III. BACKGROUND This section provide a background about the flower pollination algorithm and its multi-objective version. Section III-A introduces the flower pollination algorithm. Section III-B explains the concepts of the multi-objective optimization.

This section provide an discussion for the proposed system for EEG signals based user authentication. The proposed system run through four phases where the result of each phase is an input to the consecutive one. Phase 1 EEG signal acquisition describes in Section IV-A. Section IV-B describes Phase 2 Tuning WT parameters by using MOFPA using hybridizing between multi-objective flower pollination algorithm with wavelet transform (MOFPA-WT). Phase 3 Feature extraction from denoised EEG signals presents in Section IV-C. Phase 4 EEG signal classification using neural network classifier presents in Section IV-D. The four phases are flowcharted in Fig. 1 and thoroughly are described as follows:

A. Flower Pollination Algorithm In the recent optimization review, the meta-heuristic algorithms can be classified into: evolutionary algorithm [4], [11], swarm intelligence [9], and trajectory algorithms [2], [3], [7]. Flower pollination algorithm (FPA) is one of successful swarm-based intelligence which is inspired from the pollination behaviour of the flowering plants. FPA is introduced by Yang in 2012 [24] and successfully applied for many optimization problems [9], [17]. The rules (operators) of FPA are summarized as follows: • Rule(1): Global pollination involves the biotic and crosspollination where the pollinators are carrying the pollen based on Levy flights. • Rule(2): Local pollination involves abiotic and selfpollination. • Rule(3): The reproduction probability can be considered as the flower constancy is proportional to the similarity between any two flowers. • Rule(4): The switch probability p ∈ [0, 1] can be controlled between local pollination and global pollination Due to some external factors such as wind, local pollination will be a significant fraction p in the overall pollination activities.

Fig. 1: EEG signals based user authentication system

A. EEG signal acquisition In this study, Keirn EEG dataset has been used. More details about this dataset given in Section V. Where the original EEG signal processed using a Butterworth 5th order filter with range 6-30Hz to achieve the highest signal-to-noise ratio (SNR) and obtained the efficient features extraction. B. EEG signal denoising using MOFPA-WT

B. Multi-objective optimization •

This section describes a briefly introduction about multiobjective optimization technique. In general, the multiobjective optimization refers to solve any optimization problem using more than one objective function [25]. The multiobjective optimization problem for n objectives functions can be formulated as follows: M imimze

F (x) = f1 (x), f2 (x), ..., fn (x),

(2)

where n refers to number of objective functions. The FPA has been extended to multi-objective optimization technique by Yang et al. [25], while the author adapted multi-objective flower pollination algorithm (MOFPA) for solving engineering optimization problems. MOFPA is implemented according to the weighted sum approach to combine two objectives into a composite one objective function.

Initially, the solution of WT parameters configuration is represented as a vector Sol = (x1 , x2 , . . . xD ) where D is the total number of parameter used for WT which is equal to 5. x1 represent the value of mother wavelet function parameter Φ, x2 denotes the value of decomposition level parameter L, x3 refers to the thresholding method β, x4 represents the value of thresholding selection rule parameter λ, and x5 represents the re-scaling approach ρ, where the possible range for these parameters are selected from Table I. Fig. 2 shows the solution of WT parameters for EEG signal denoising. The proposed method MOFPAWT evaluates the solution using the multi-objective function which is formulated in Eq.(3). MOFPA-WT applied using two objective functions manly: min(MSE) and max(SNR) to achieve the best combination of WT parameters for EEG signal denoising [10].

Fig. 2: Solution of WT parameters for denoising EEG signals

that can be extracted from the denoised EEG signal. In this paper, we applied we have used four popular measurements of the signal which are mean, standard deviation, entropy, and energy where these features are able to provide a unique patterns among the users. These four features are formulated as follows: N

EEGM ean FM OF P AW T = (W1 ∗min(M SE))+(W2 ∗max(SN R)) (3) N X 1 M SE = [x(n) − x b(n)]2 (4) N n=1 

PN

SN Rout = 10 log10 PN

n=1 [x(n)]

n=1 [x(n)

•

2

−x b(n)]2



(5)

where x(n) denotes the original EEG signal and x b(n) is the denoised EEG signal obtained by tuning the wavelet parameters using MOFPA-WT. Iteratively, the randomly generated solutions undergoes refinement using the MOFPA-WT. The final result of this 0 phase is an optimized solution Solopt = (x01 , x02 , . . . x0D ) which will be passed to the next phase. 0 . As aforeEEG denoising using WT based on Solopt mentioned in Section II, the denoising process of WT involves three main steps that are visualized in Figure 3 and described in more details below: – EEG signal decomposition using DWT. In this step the DWT is applied to decompose the noise of the input EEG signals x(n). In decomposition process, 0 parameters, namely, we must use the first two Solopt the mother wavelet furcation ρ and the decomposition level L). Figure 3 shows the DWT procedure for five levels, where the input EEG signal is divided at each level into cA and cD. The latter is processed using a high-pass filter, while the former is processed using a low-pass filter and is decomposed for the next level. – The second step of EEG denoising is Thresholding which is applied based on the noise level of the coefficients. In this step, the last three wavelet parameters, namely, the thresholding type (β), the thresholding selection rules (λ), and the re-scaling 0 methods (ρ), must be selected from Solopt .

C. Feature Extraction Extracting efficient features considers a significant phase in any authentication system because it will increase the performance of the proposed system to get good results in the correct classification [21], [23]. Therefore, the main purpose of this phase is to find the unique characteristics features from each sub-bands (i.e., high gamma, gamma, alpha, beta, theta, and delta). Fig. 3 shows feature extraction based WT decomposition with five levels. There are several features

EEGStd

1 X ∗ Dij , i = 1, 2, 3, ..., L, = N j=1

v u N u1 X (xi − x)2 , i = 1, 2, 3, ..., L =t N j=i

EEGEnergy =

N X

| Dij |2 , i = 1, 2, 3, ..., L

(6)

(7)

(8)

j=1

EEGEntropy = −

X

p(x) log p(x),

(9)

Fig. 3: EEG feature extraction based on WT 5 decomposition level Table II shows the total number of features which are used in this paper. D. Neural Network classifier To classify the extracted features from the denoised EEG signal into correct person artificial neural network (ANN) classifier has been applied. We used ANN and pattern recognition tool for classification task and designed a network with 24input features vector of each subject (i. e., 4 features * 6 subbands) and 32 hidden layers and 7 output layers because we tested the proposed system on seven users. V. R ESULTS AND D ISCUSSIONS Keirn EEG dataset has been used in this paper. This dataset recorded EEG from subjects while they were performing different mental tasks exploring new human-machine interaction through the brain [14]. Although a small database (7

TABLE II: Total No. of features the EEG dataset No. of subjects (X) 7

No. of tasks (S) 5

No. of trails (Tr) 10

No. of Session (Ta) 2

subjects: males and females between the ages of 21 and 48) the relevance of this database resides on the multi-task recording paradigm. In particular, a total of 5 tasks were performed by subjects. Each task was repeated 5 times and recorded under both Rest Eyes Closed (REC) and Rest Eyes Open (REO) on every session. 2 sessions were recorded from each subject in a time span of 2 weeks. The tasks were: Task 1 Base line measurements. This task was taken as a baseline for comparison. In this case, subjects were only asked to relax. Task 2 Complex problem solving. Subjects were asked to mentally solve non-trivial multiplication problems. Task 3 Geometric figure rotation. Subjects were presented with an image of a 3 dimensional complex object before being asked to mentally rotate it. Task 4 Mental letter composition. Subjects had to mentally write a letter to a friend or a family member. Task 5 Visual counting. Subjects were asked to visualize numbers being written on a blackboard sequentially. With the previous number being erased before a new number is written. The EEG dataset which used in this paper have been separated into five different mental tasks based on 10folds cross-validation technique for training and testing for each task. To evaluate the performance of the MOFPA-WT method three measures have been calculated namely, accuracy, true acceptance rate (TAR), and false acceptance rate (FAR) which can formulated as follows: TA + TR (10) Accuracy = TA + FA + TR + FR

No. sub-band 6

no. of features 4

Total No. of features 16,800 features

TABLE III: Confusion matrix dataset of baseline task Sub1 Sub2 Sub3 Sub4 Sub5 Sub6 Sub7 FAR

Sub1

Sub2

Sub3

Sub4

Sub5

Sub6

Sub7

Spe

Sen

50 0 0 8 0 0 1 0.027

0 15 1 0 1 7 1 0.028

0 2 51 1 1 2 1 0.021

9 2 2 44 2 0 7 0.067

1 3 0 3 80 7 0 0.047

0 8 6 0 5 43 0 0.058

0 0 0 4 1 1 20 0.017

0.973 0.972 0.979 0.933 0.953 0.942 0.983

0.833 0.5 0.85 0.733 0.889 0.717 0.667

Spe is Specificity and Sen is Sensitivity. Accuracy=77.69% and FAR=22.31%

TABLE IV: Confusion matrix dataset of multiplication task Sub1 Sub2 Sub3 Sub4 Sub5 Sub6 Sub7 FAR

Sub1

Sub2

Sub3

Sub4

Sub5

Sub6

Sub7

Spe

Sen

51 0 0 7 1 0 1 0.027

0 22 2 1 1 5 0 0.025

0 3 55 2 0 0 0 0.015

8 2 1 39 7 0 0 0.055

1 1 0 9 78 3 0 0.047

0 2 0 1 3 52 0 0.018

0 0 2 1 0 0 29 0.008

0.973 0.975 0.985 0.945 0.953 0.982 0.992

0.85 0.733 0.917 0.65 0.867 0.867 0.967

Spe is Specificity and Sen is Sensitivity. Accuracy=83.58% and FAR=16.42%

TABLE V: Confusion matrix dataset of rotation task Sub1 Sub2 Sub3 Sub4 Sub5 Sub6 Sub7 FAR

Sub1

Sub2

Sub3

Sub4

Sub5

Sub6

Sub7

Spe

Sen

46 0 0 8 0 0 1 0.027

0 21 3 0 1 4 0 0.019

0 3 57 0 0 0 0 0.009

12 0 0 49 3 0 0 0.045

1 2 0 3 86 2 0 0.027

0 4 1 0 0 54 0 0.015

1 0 0 0 0 0 29 0.003

0.973 0.981 0.991 0.955 0.973 0.985 0.997

0.767 0.7 0.95 0.817 0.956 0.9 0.967

Spe is Specificity and Sen is Sensitivity. Accuracy=87.69% and FAR=12.31%

TA Sensitivity(T AR) = TA + FR

(11)

TR Specif ity(T F R) = TR + FR

(12)

F AR = 1 − T F R

(13)

TABLE VI: Confusion matrix dataset of letter composing task

where TA,TR,FA, and FR represent true acceptance, true reject, false acceptance, and false reject, respectively. The results of classification phase are represented as a confusion matrix that tabulates whether they fall into one of four categories: true acceptance (TA), true reject (TR), false acceptance (FA) and false reject (FR). Tables (III, IV, V,VII, and VI) show the confusion matrixes according to base line, multiplication, rotation, letter composition, and counting task, respectively. Overall, subject seven obtains the best results during the test for all tasks where it achieved the highest sensitivity value=1 with counting task. The highest accuracy is obtained with rotation task, while the accuracy=87.69%, and FAR= 12.31%. Figure 4 shows the accuracy rate for the input EEG signals based on five decomposition Level using MOFPA-WT for base line,

Sub1 Sub2 Sub3 Sub4 Sub5 Sub6 Sub7 FAR

Sub1

Sub2

Sub3

Sub4

Sub5

Sub6

Sub7

Spe

Sen

57 0 0 2 0 0 2 0.012

0 18 0 3 1 6 1 0.031

0 0 57 0 0 1 1 0.006

2 4 0 42 2 4 1 0.039

0 0 0 2 84 3 2 0.024

0 7 2 3 3 46 0 0.046

1 1 1 2 0 0 23 0.014

0.95 0.6 0.95 0.778 0.933 0.767 0.767

0.988 0.969 0.994 0.961 0.976 0.954 0.986

Spe is Specificity and Sen is Sensitivity. Accuracy=85.15% and FAR=14.85%

TABLE VII: Confusion matrix dataset of counting task Sub1 Sub2 Sub3 Sub4 Sub5 Sub6 Sub7 FAR

Sub1

Sub2

Sub3

Sub4

Sub5

Sub6

Sub7

Spe

Sen

50 0 0 19 0 0 0 0.058

0 25 5 0 0 2 0 0.019

0 3 52 0 0 0 0 0.009

9 0 0 33 5 0 0 0.042

1 1 0 6 85 1 0 0.03

0 1 1 0 0 57 0 0.006

0 0 2 2 0 0 30 0.011

0.942 0.981 0.991 0.958 0.97 0.994 0.989

0.833 0.833 0.867 0.55 0.944 0.95 1

Spe is Specificity and Sen is Sensitivity. Accuracy=85.13% and FAR=14.87%

multiplication, rotation, letter composition, and counting task, respectively.

Fig. 4: Accuracy results based on five decomposition Level using MOFPA-WT VI. C ONCLUSION AND FUTURE WORK In this paper a novel technique for EEG signal denoising based on multi-objective flower pollination algorithm with wavelet transform (MOFPA-WT) is proposed. The main task of MOFPA-WT method is to find the efficient decomposition of the input EEG signal which can provide a unique features from each sub-bands. MOFPA-WT is tested using a standard EEG signal dataset, namely, Keirn EEG dataset which have five mental tasks includes base line, multiplication, rotation, letter composition, and counting task, respectively. The performance of MOFPA-WT is evaluated using three criteria, namely, accuracy, TAR, and FAR. In conclusion, the proposed method achieves the highest accuracy result which can be obtained using mental tasks based on geometric figure rotation compared with mental tasks. For future work, MOFPA-WT will be applied for more challenging signal problem instances, such as user authentication with large EEG dataset or early detection of epilepsy based on EEG signal. Furthermore, the real-world applications are required to show the efficiency of MOFPA-WT. Other possible improvements are applying one of features selection technique is recommended to increase the accuracy rate as well as to reduce the dimensions redundancy of the extracted EEG features. ACKNOWLEDGMENT The first author would like to thank the University Science Malaysia (USM) and The World Academic Science (TWAS) for supporting his PhD study which is USM-TWAS Postgraduate Fellowship, FR number: 3240287134. R EFERENCES [1] S. N. Abdulkader, A. Atia, and M.-S. M. Mostafa. Brain computer interfacing: Applications and challenges. Egyptian Informatics Journal, 16(2):213–230, 2015. [2] L. M. Abualigah, A. T. Khader, M. A. Al-Betar, Z. A. A. Alyasseri, O. A. Alomari, and E. S. Hanandeh. Feature selection with β-hill climbing search for text clustering application. In Information and Communication Technology (PICICT), 2017 Palestinian International Conference on, pages 22–27. IEEE, 2017. [3] M. A. Al-Betar. β -hill climbing: an exploratory local search. Neural Computing and Applications, pages 1–16, 2016.

[4] M. A. Al-Betar, Z. A. A. Alyasseri, A. T. Khader, A. L. Bolaji, and M. A. Awadallah. Gray image enhancement using harmony search. International Journal of Computational Intelligence Systems, 9(5):932– 944, 2016. [5] Z. A. A. Alyasseri, A. T. Khader, and M. A. Al-Betar. Electroencephalogram signals denoising using various mother wavelet functions: A comparative analysis. In ICISPC ’17, Penang, Malaysia, pages 66–70. ACM, 2017. [6] Z. A. A. Alyasseri, A. T. Khader, and M. A. Al-Betar. Optimal eeg signals denoising using hybrid β-hill climbing algorithm and wavelet transform. In ICISPC ’17, Penang, Malaysia, pages 147–152. ACM, 2017. [7] Z. A. A. Alyasseri, A. T. Khader, M. A. Al-Betar, and L. M. Abualigah. Ecg signal denoising using β-hill climbing algorithm and wavelet transform. In ICIT 2017 The 8th International Conference on Information Technology, pages 1–7, 2017. [8] Z. A. A. Alyasseri, A. T. Khader, M. A. Al-Betar, and M. A. Awadallah. Hybridizing β-hill climbing with wavelet transform for denoising ecg signals. Information Sciences, 429:229–246, 2018. [9] Z. A. A. Alyasseri, A. T. Khader, M. A. Al-Betar, M. A. Awadallah, and X.-S. Yang. Variants of the flower pollination algorithm: A review. In Nature-Inspired Algorithms and Applied Optimization, pages 91–118. Springer, 2018. [10] Z. A. A. Alyasseri, A. T. Khader, M. A. Al-Betar, P. P. Joao, and A. A. Osama. Eeg-based person authentication using multi-objective flower pollination algorithm. In Evolutionary Computation (CEC), 2018 IEEE Congress on, pages 15–22. IEEE, 2018. [11] Z. A. A. Alyasseri, I. Venkat, M. A. Al-Betar, and A. T. Khader. Edge preserving image enhancement via harmony search algorithm. In Data Mining and Optimization (DMO), 2012 4th Conference on, pages 47–52. IEEE, 2012. ¨ [12] H. Berger. Uber das elektrenkephalogramm des menschen. European Archives of Psychiatry and Clinical Neuroscience, 87(1):527–570, 1929. [13] D. L. Donoho and J. M. Johnstone. Ideal spatial adaptation by wavelet shrinkage. Biometrika, 81(3):425–455, 1994. [14] Z. A. Keirn and J. I. Aunon. A new mode of communication between man and his surroundings. IEEE transactions on biomedical engineering, 37(12):1209–1214, 1990. [15] P. Kumari and A. Vaish. Brainwave based user identification system: A pilot study in robotics environment. Robotics and Autonomous Systems, 65:15–23, 2015. [16] P. Kumari and A. Vaish. Feature-level fusion of mental tasks brain signal for an efficient identification system. Neural Computing and Applications, 27(3):659–669, 2016. [17] A. A. Osama, A. T. Khader, M. A. Al-Betar, and Z. A. A. Alyasseri. A hybrid filter-wrapper gene selection method for cancer classification. In 2nd International Conference on BioSignal Analysis, Processing and Systems (ICBAPS), pages 32–37. IEEE, 2018. [18] R. A. Ramadan and A. V. Vasilakos. Brain computer interface: control signals review. Neurocomputing, 223:26–44, 2017. [19] R. P. Rao. Brain-computer interfacing: an introduction. Cambridge University Press, 2013. [20] D. Rodrigues, G. F. Silva, J. P. Papa, A. N. Marana, and X.-S. Yang. Eegbased person identification through binary flower pollination algorithm. Expert Systems with Applications, 62:81–90, 2016. [21] N. D. Sarier. Improving the accuracy and storage cost in biometric remote authentication schemes. Journal of Network and Computer Applications, 33(3):268–274, 2010. [22] G. Schalk, D. J. McFarland, T. Hinterberger, N. Birbaumer, and J. R. Wolpaw. Bci2000: a general-purpose brain-computer interface (bci) system. IEEE Transactions on biomedical engineering, 51(6):1034– 1043, 2004. [23] P. K. Sharma and A. Vaish. Individual identification based on neurosignal using motor movement and imaginary cognitive process. OptikInternational Journal for Light and Electron Optics, 127(4):2143–2148, 2016. [24] X.-S. Yang. Flower pollination algorithm for global optimization. In International Conference on Unconventional Computing and Natural Computation, pages 240–249. Springer, 2012. [25] X.-S. Yang, M. Karamanoglu, and X. He. Flower pollination algorithm: a novel approach for multiobjective optimization. Engineering Optimization, 46(9):1222–1237, 2014.