Epilepsy diagnosis using artificial neural network learned by PSO$^{\dag}$

Yalçin, Nesibe; Tezel, Gülay; Karakuzu, Cihan

doi:10.3906/elk-1212-151

Cited by 38 publications

(17 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this way, some specific latent features which characterize the dynamical and nonlinear frameworks in the signals can be obtained from these sub-transformed bands. [9][10][11][12][13][14] In the literature, there are remarkable approaches in which different feature extraction techniques are processed to extract different types of features such as the statistical moments, entropy and metric measures. [15][16][17][18] In the context of seizure detection, the most commonly used techniques are Discriminant Analysis, Logistic Regression, Gaussian Mixture Models, Regression Trees, Random Forest, k-Nearest Neighbor, Naive Bayes Classifier, Kernel Methods, Support Vector Machines (SVMs), ANNs and ANFIS classifiers.…”

mentioning

confidence: 99%

“…Especially, ANN classifiers are often used together with WTs due to their flexible and adjustable structures. 10,11,14,[19][20][21][22][23][24][25][26][27] Also, there are many studies in which different feature extraction techniques are perefered with the other classifiers rather than ANNs. 1,12,13,[15][16][17][28][29][30][31][32][33][34][35][36][37][38][39][40]…”

mentioning

confidence: 99%

See 1 more Smart Citation

Classification of EEG Signals for Epileptic Seizures Using Linear and Non-linear Classifiers Based Wavelet Transforms and Information Criteria

Özer¹,

Kocadağlı²

2019

Turkiye Klinikleri J Biostat

View full text Add to dashboard Cite

Objective: This study presents an efficient procedure that provides an accurate classification of Electroencephalogram (EEG) signals for the detection of epileptic seizure. Essentially, the proposed procedure hybridizes the linear and nonlinear classifiers with the discrete wavelet transforms (DWT) and principal component analysis (PCA), separately. Material and Methods: To classify EEG signals more accurately, the proposed multi-resolution signal processing technique splits them into the detailed partitions with different window-widths, and then decomposes them into detail and approximation coefficients by means of DWT. Thus, many specific latent features that characterize the nonlinear and dynamical structures in the signals can be evaluated from these coefficients. During the model estimation process with multivariate logistic regression (MLR) and artificial neural networks (ANNs), to control the complexity of model and reduce the dimension of feature matrix, PCA is used. In addition, to quantify the complexity and select the best models, the information criteria are considered for both MLR and ANNs. To improve the classification performance, ANNs are trained by various gradient algorithms as well as considering early stopping and cross-validation techniques. Results: According to analysis results over the benchmark epilepsy data set released by the Department of Epileptology at University of Bonn, the proposed approach is to bring out 99% accuracy ratios for classifying the epileptic signals. Conclusion: This approach not only allows making an efficient analysis of EEG signals for detection of epilepsy, but also provides the best model configurations for ANNs and MLR in terms of reliability and complexity.Keywords: EEG signal processing; epileptic seizures; discrete wavelet transform; artificial neural networks; multinomial logistic regression; principal component analysis ÖZET Amaç: Bu çalışma, epileptik nöbetlerin tesbiti için Elektroensefalogram (EEG) sinyallerini doğru sınıflandıran etkin bir yöntem önermektedir. Esas olarak, bu yöntem lineer ve lineer olmayan sınıflandırıcıları, ayrık dalgacık dönüşümleri (ADD) ve temel bileşenler analizi (TBA) ile hibritleştirmektedir. Gereç ve Yöntemler: Önerilen çoklu-çözünürlüklü sinyal işleme tekniği, EEG sinyallerinin daha doğru sınıflandırılmak için onları farklı bant genişlikli parçalara bölmekte ve bu parçaları ADD yardımıyla ayrıntı (detail) ve yaklaşım (approximate) katsayılarına ayrıştırmaktadır. Böylece, sinyallerin barındırdığı dinamik ve lineer olmayan yapıları karaterize eden birçok gizli özellik, bu katsayılar üzerinden belirlenmektedir. Çokterimli Logistik Regregresyon (ÇLR) ve Yapay Sinir Ağlarıyla (YSA) model kestirim sürecinde, karmaşıklığı kontrol etmek ve veri matrisini indirgemek için TBA kullanılmıştır. Bunun yanısıra, model karmaşıklığının nicelendirilmesi ve en iyi modellerin belirlenmesi için bilgi kriterlerinden yararlanılmıştır. Doğru sınıflandırma performasını arttırmak için YSA'lar erken durdurma ve çapraz geçerlilik teknikleriyle beraber ...

show abstract

mentioning

confidence: 99%

mentioning

confidence: 99%

Classification of EEG Signals for Epileptic Seizures Using Linear and Non-linear Classifiers Based Wavelet Transforms and Information Criteria

Özer¹,

Kocadağlı²

2019

Turkiye Klinikleri J Biostat

View full text Add to dashboard Cite

show abstract

“…Calculate fitness value of the particle if fitness value of the current particle < fitness value of the pbest particle then The velocity and position updating of a particle at k th generation [22,23].…”

Section: Do For Each Particle Domentioning

confidence: 99%

Bandwidth extension of narrowband speech in log spectra domain using neural network

Pourmohammadi¹,

Vali²,

Ghadyani³

2015

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Electroencephalogram (EEG) is used routinely for diagnosis of diseases occurring in the brain. It is a very useful clinical tool in the classification of epileptic seizures and the diagnosis of epilepsy. In this study, epilepsy diagnosis has been investigated using EEG records. For this purpose, an artificial neural network (ANN), widely used and known as an active classification technique, is applied. The particle swarm optimization (PSO) method, which does not need gradient calculation, derivative information, or any solution of differential equations, is preferred as the training algorithm for the ANN. A PSO-based neural network (PSONN) model is diversified according to PSO versions, and 7 PSO-based neural network models are described. Among these models, PSONN3 and PSONN4 are determined to be appropriate models for epilepsy diagnosis due to having better classification accuracy. The training methods-based PSO versions are compared with the backpropagation algorithm, which is a traditional method. In addition, different numbers of neurons, iterations/generations, and swarm sizes have been considered and tried. Results obtained from the models are evaluated, interpreted, and compared with the results of earlier works done with the same dataset in the literature.

show abstract

“…As a result, to obtain optimum SVM parameters, various optimization techniques have been used such as the grid search method (GSM) [20], genetic algorithm (GA) [21,22], and particle swarm optimization (PSO) [23,24]. Although these methods are quite effective, they suffer from becoming trapped into local optima and from excessive time requirements.…”

Section: Introductionmentioning

confidence: 99%

A novel perturbed particle swarm optimization-based support vector machine for fault diagnosis in power distribution systems

Thom¹,

Cho²,

Tuan³

2018

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Abstract:In this paper, a novel perturbed particle swarm optimization (PPSO) algorithm is investigated to improve the performance of a support vector machine (SVM) for short-circuit fault diagnosis in power distribution systems.In the proposed PPSO algorithm, the velocity of each particle is perturbed whenever the particles strike into a local optimum, in order to achieve a higher quality solution to optimization problems. Furthermore, the concept of proposed perturbation is applied to three variants of PSO, and improved corresponding algorithms are named perturbed C-PSO (PC-PSO), perturbed T-PSO (PT-PSO), and perturbed K-PSO (PK-PSO). For the purpose of fault diagnosis, the timedomain reflectometry (TDR) method with pseudorandom binary sequence (PRBS) excitation is considered to generate the necessary fault simulation data set. The proposed approaches are tested on a typical two-lateral radial distribution network.

show abstract

Epilepsy diagnosis using artificial neural network learned by PSO$^{\dag}$

Cited by 38 publications

References 19 publications

Classification of EEG Signals for Epileptic Seizures Using Linear and Non-linear Classifiers Based Wavelet Transforms and Information Criteria

Classification of EEG Signals for Epileptic Seizures Using Linear and Non-linear Classifiers Based Wavelet Transforms and Information Criteria

Bandwidth extension of narrowband speech in log spectra domain using neural network

A novel perturbed particle swarm optimization-based support vector machine for fault diagnosis in power distribution systems

Contact Info

Product

Resources

About