Ömer Faruk Alçin

A hybrid method based on time–frequency images for classification of alcohol and control EEG signals

Classification of alcoholic electroencephalogram (EEG) signals is a challenging job in biomedical research for diagnosis and treatment of brain diseases of alcoholic people. The aim of this study was to introduce a robust method that can automatically identify alcoholic EEG signals based on time–frequency (T–F) image information as they convey key characteristics of EEG signals. In this paper, we propose a new hybrid method to classify automatically the alcoholic and control EEG signals. The proposed scheme is based on time–frequency images, texture image feature extraction and nonnegative least squares classifier (NNLS). In T–F analysis, the spectrogram of the short-time Fourier transform is considered. The obtained T–F images are then converted into 8-bit grayscale images. Co-occurrence of the histograms of oriented gradients (CoHOG) and Eig(Hess)-CoHOG features are extracted from T–F images. Finally, obtained features are fed into NNLS classifier as input for classify alcoholic and control EEG signals. To verify the effectiveness of the proposed approach, we replace the NNLS classifier by artificial neural networks, k-nearest neighbor, linear discriminant analysis and support vector machine classifier separately, with the same features. Experimental outcomes along with comparative evaluations with the state-of-the-art algorithms manifest that the proposed method outperforms competing algorithms. The experimental outcomes are promising, and it can be anticipated that upon its implementation in clinical practice, the proposed scheme will alleviate the onus of the physicians and expedite neurological diseases diagnosis and research.

OMP-ELM: Orthogonal Matching Pursuit-Based Extreme Learning Machine for Regression

Abstract Extreme learning machine (ELM) is a recent scheme for single hidden layer feed forward networks (SLFNs). It has attracted much interest in the machine intelligence and pattern recognition fields with numerous real-world applications. The ELM structure has several advantages, such as its adaptability to various problems with a rapid learning rate and low computational cost. However, it has shortcomings in the following aspects. First, it suffers from the irrelevant variables in the input data set. Second, choosing the optimal number of neurons in the hidden layer is not well defined. In case the hidden nodes are greater than the training data, the ELM may encounter the singularity problem, and its solution may become unstable. To overcome these limitations, several methods have been proposed within the regularization framework. In this article, we considered a greedy method for sparse approximation of the output weight vector of the ELM network. More specifically, the orthogonal matching pursuit (OMP) algorithm is embedded to the ELM. This new technique is named OMP-ELM. OMP-ELM has several advantages over regularized ELM methods, such as lower complexity and immunity to the singularity problem. Experimental works on nine commonly used regression problems indicate that the investigated OMP-ELM method confirms these advantages. Moreover, OMP-ELM is compared with the ELM method, the regularized ELM scheme, and artificial neural networks.

Extreme learning machine based robotic arm modeling

Robotic arms are very powerful machines that can be used in many various applications in industry. So that, a suitable dynamic model is derived to verify that performs the tasks. But, dynamic equation is an important issue due to its complexity. Thus, an alternative model can be derived for the robotic arms. This paper is proposed Extreme Learning Machine (ELM) model for the angular acceleration of a robotic arm. The performance of the ELM model is performed by using Pumadyn datasets. At the same time, the validation of the proposed model is compared with Artificial Neural Network (ANN). Experimental results show that the proposed model is suitable and it provides low computation complexity.

Analysis of complex extreme learning machine-based nonlinear equalizer for coherent optical OFDM systems

One major drawback of coherent optical OFDM (CO-OFDM) is its vulnerability to nonlinear fiber effects due to its high peak-to-average power ratio. Fiber nonlinearities can be mitigated using machine learning algorithms that are a nonlinear decision classifier. In this study, C-ELM based nonlinear equalizer is proposed for a MQAM CO-OFDM. MQAM CO-OFDM systems are simulated by designing a Monte Carlo simulation. In this simulation, the effect of fiber nonlinearities on received signals is demonstrated with constellation diagrams and results are given in form of BER-Fiber Length variations.

Hilbert transform based simple detection and indice analyze of voltage sags using synthetic data

Electrical grid has lots of changes in its morphology and managing style since first installed nearly two hundred years ago. Today, smart grid structure plays a crucial role when creating a sustainable and reliable operation. In smart grid context, power quality issues are monitored and required measures are obtained from smart meters. Power quality term include voltage quality. When it is about voltage quality, sags take the lead among other disturbances. System operators have to track voltage sags to provide a better service quality. In this study, a fast and simple algorithm called Hilbert Transform is used to detect voltage sags in synthetic dataset. Then, a voltage sag table is built considering related IEEE and IEC standards to identify site indices SARFI-X and SIARFI-X. Purpose of the study is being a first step to voltage sag detection and defining indices with real data. Obtained results denote and feed this aim.

Machine learning based power quality event classification using wavelet — Entropy and basic statistical features

Todays industrial environment is smarter than ever before. Most production lines include electrical devices which are able to communicate each other and controlled from a single station with automation systems. Most of those elements have an internet connection link known as industrial internet. Development of smart technology with industrial internet comes with a need of monitoring. Monitoring technologies are emergent systems that focus on fault detection, grid self - healings and online tracking of power quality issues. Present study deals with one of the essential part of an electricity grid monitoring system called power quality event classification in a manner of machine learning topic. Power quality events to be processed are generated synthetically by means of a comprehensive software tool. Classification of real-like dataset is executed using extreme learning machine which is an extremely fast learning algorithm applied to single layer neural networks. Basic statistical criteria and wavelet - entropy methods are handled to achieve distinctive features of dataset. As a performance evaluation instrument, conventional artificial neural network structure is run too. Detailed results are discussed to prove the satisfactory performance of proposed pattern recognition model.

Bir Boyutlu Yerel İkili Örüntüler ve Ayrık Dalgacık Dönüşümü Tabanlı Yeni Bir Güç Kalitesi Olay Sınıflandırma Yöntemi

Bu makalede, gerilim cokmesi, yukselmesi ve kesintisi, gerilim harmonikleri ve gecici durumlardan olusan guc kalitesi bozulmalarina ait olay verilerini siniflandirmak icin akilli bir oruntu tanima sistemi incelenmistir. Onerilen sistemin altyapisini, oznitelik cikarimi ve siniflandirma asamalari olusturmaktadir. Ayirt edici ozniteliklerin cikarilmasi islemi siniflandirici performansini etkileyen en onemli unsurlar arasinda yer almaktadir. Onerilen calismada Ayrik Dalgacik Donusumu (ADD) ve Bir Boyutlu Yerel Ikili Oruntu (1B-YIO) yontemlerinden elde edilen oznitelikler kullanilmistir. Guc Kalitesi Olay (GKO) siniflandirma isleminde daha once incelenmemis yeni bir yontem olan 1B-YIO yontemi, ADD ozellikleri ile birlikte ele alinarak siniflandirici basarimi incelenmistir. Veri setini olusturan GKO isaretleri kapsamli bir yazilim araci ile uretilmistir. Matematiksel modeller kullanilarak olusturulan bu aracta GKO verilerini iceren veri seti gercege en yakin haliyle elde edilmistir. Siniflandirici olarak bircok uygulamada yaygin olarak kullanilan Uc Ogrenme Makinesi (UOM) tercih edilmistir. Basarim degerlendirmesinin etkinligini artirmak icin geleneksel Yapay Sinir Agi (YSA) tabanli siniflandiriciya ait sonuclar da elde edilmistir. Cesitli gurultu icerigi de dikkate alinarak yapilan deneysel calismalarda siniflandirici basariminin kabul edilebilir degerlere ulastigi kaydedilmistir. Calismaya ait sonuclar detayli olarak gosterilmistir.

A Novel Edge Detection Algorithm Based on Texture Feature Coding

Abstract A new edge detection technique based on the texture feature coding method (TFCM) is proposed. The TFCM is a texture analysis scheme that is generally used in texture-based image segmentation and classification applications. The TFCM transforms an input image into a texture feature image whose pixel values represent the texture information of the pixel in the original image. Then, on the basis of the transformed image, several features are calculated as texture descriptors. In this article, the TFCM is employed differently to construct an edge detector. In particular, the texture feature number (TFN) of the TFCM is considered. In other words, the TFN image is used for subsequent processes. After obtaining the TFN image, a simple thresholding scheme is employed for obtaining the coarse edge image. Finally, an edge-thinning procedure is used to obtain the tuned edges. We conducted several experiments on a variety of images and compared the results with the popular existing methods such as the Sobel, Prewitt, Canny, and Canny–Deriche edge detectors. The obtained results were evaluated quantitatively with the Figure of Merit criterion. The experimental results demonstrated that our proposed method improved the edge detection performance greatly. We further implemented the proposed edge detector with a hardware system. To this end, a field programmable gate array chip was used. The related simulations were carried out with the MATLAB Simulink tool. Both software and hardware implementations demonstrated the efficiency of the proposed edge detector.