Ilyas Eminoglu

Reviewing and designing pre-processing units for RBF networks: initial structure identification and coarse-tuning of free parameters

This paper reviews some frequently used methods to initialize an radial basis function (RBF) network and presents systematic design procedures for pre-processing unit(s) to initialize RBF network from available input–output data sets. The pre-processing units are computationally hybrid two-step training algorithms that can be named as (1) construction of initial structure and (2) coarse-tuning of free parameters. The first step, the number, and the locations of the initial centers of RBF network can be determined. Thus, an orthogonal least squares algorithm and a modified counter propagation network can be employed for this purpose. In the second step, a coarse-tuning of free parameters is achieved by using clustering procedures. Thus, the Gustafson–Kessel and the fuzzy C-means clustering methods are evaluated for the coarse-tuning. The first two-step behaves like a pre-processing unit for the last stage (or fine-tuning stage—a gradient descent algorithm). The initialization ability of the proposed four pre-processing units (modular combination of the existing methods) is compared with three non-linear benchmarks in terms of root mean square errors. Finally, the proposed hybrid pre-processing units may initialize a fairly accurate, IF–THEN-wise readable initial model automatically and efficiently with a minimum user inference.

DOPGA: a new fitness assignment scheme for multi-objective evolutionary algorithms

In this article, a new fitness assignment scheme to evaluate the Pareto-optimal solutions for multi-objective evolutionary algorithms is proposed. The proposed DOmination Power of an individual Genetic Algorithm (DOPGA) method can order the individuals in a form in which each individual (the so-called solution) could have a unique rank. With this new method, a multi-objective problem can be treated as if it were a single-objective problem without drastically deviating from the Pareto definition. In DOPGA, relative position of a solution is embedded into the fitness assignment procedures. We compare the performance of the algorithm with two benchmark evolutionary algorithms (Strength Pareto Evolutionary Algorithm (SPEA) and Strength Pareto Evolutionary Algorithm 2 (SPEA2)) on 12 unconstrained bi-objective and one tri-objective test problems. DOPGA significantly outperforms SPEA on all test problems. DOPGA performs better than SPEA2 in terms of convergence metric on all test problems. Also, Pareto-optimal solutions found by DOPGA spread better than SPEA2 on eight of 13 test problems.

Classification of two common power quality disturbances using wavelet based SVM

Development of technology increased the attention of the research community on power quality (PQ) disturbance classification problem. This paper presents wavelet based effective feature extraction method and support vector machines (SVM) for PQ disturbance classification problem. Two common kinds of power quality disturbances, voltage sag and swell, are considered in this paper. After multi-resolution signal decomposition of PQ disturbances, feature vector can be obtained. Multi-resolution analysis (MRA) technique of discrete wavelet technique (DWT) and Parsevals theorem are employed to extract the energy distribution features of sag and swell signals. SVM are used to classify these feature vectors of PQ disturbances. Performance of two kinds of method used in SVM is compared aspect of training time and training error.

Neuromuscular disease diagnosis of SVM, K-NN and DA algorithm based classification part-II

This study includes a classification structure consisting of second part for the automatic diagnosis of the neuromuscular disease of ALS (Amyotrophic Lateral Sclerosis) and myopathy being a muscular disease. In this study feature vectors containing time domain parameters, frequency domain parameters (a total of 25 feature vectors) as well as feature vectors composed of combination of these parameters were used. In the classification stage, Support Vector Machines (SVM), K-Nearest Neighbors (K-NN) and Discriminant Analysis (DA) algorithms were employed. Experimental results showed that the multiple feature vectors proved to be more successful compared to the individual feature vectors. It is understood with this study; the classification performance depends highly on separability of feature vectors between different classes.

Classification of ALS disease using support vector machines

In this study, SVM (Support Vector Machine) algorithm is used for the diagnosis of ALS which is the most common type of motor neuron disease. Before classification of EMG data with SVM (Support Vector Machine); pre-processing, segmentation, feature extraction and clustering stages of data are completed. In the stage of clustering, hybrid and hierarchical clustering methods are employed. After that, feature vectors in time and frequency domains and their different combinations (a total of 11 feature vectors) are fed to the SVM and the obtained results are observed. It is understood that the advantages of clustering methods dependent on the feature vectors; multiple feature vectors provide high performance in the diagnosis of ALS disease and exhibit much lower discrepancy.

Fuzzy logic based classification for multifunctional upper limb prostheses

This paper investigate a fuzzy logic based high performance surface EMG classification algorithm for multifunctional upper limb prostheses. In this paper, we record 4 channels EMG data with surface electrodes from the forearm and aimed to recognize 8 different upper limb movements using heuristic fuzzy logic methods from these data. We use 50 surface EMG data for every function and evaluate the performance of this method.

Multifunctional hand prosthesis setup design

In this study, two channels, multifunctional myoelectric prosthesis hand is designed and operated for biomedical and control engineering departments laboratory as experiment apparatus. Various analog signal processing steps are implemented to the surface EMG signals which received on the triceps and biceps muscles. After that signals are conveyed to microcontroller and compared with a threshold value. One function of five different functions is selected by the received signal from triceps muscles and this function is activated by the biceps muscles. The goal of this study is, which includes different types of engineering issues, improving the theoretical and practical knowledge of engineering students.

Programmable current source and nerve conduction velocity experimental setup

In this study, programmable current source (PCS) experimental set-up that measures the nerve conduction velocity is designed. Firstly, analog PCS is implemented with integrated circuit of 555. Then, microprocessor-based PCS is designed by improving existing analog design at hand. PCS can produce current signal with adjustable frequency and amplitude. An EMG amplifier circuit is designed to measure the EMG signal from thumb finger muscle stimulated by PCS device. EMG signal is then send to the computer with a serial communication port. Received EMG signals are plotted in MATLAB and some interpretations are given.

Design of prosthetic hand simulator

In this study, Virtual Prosthetic Hand that controlled by single and dual channel sEMG (surface EMG) signals is presented. Virtual hand is designed by Blender 3D that is a powerful graphic modelling tool. Additionally, processing of raw EMG signal is explained with MATLAB. Python software language which has a large library was used for importing and evaluating of signals and for control algorithms. The Presented virtual hand has a modular structure and can be controlled by single and dual sEMG signals. Virtual hand opens, closes at constant speed and remains its position according to the processing signals.

Estimate angle information of hand open-close from surface electromyogram (sEMG)

In this paper, an estimation of angle of hand opening-closing movements by using the Artificial Neural Network (ANN) from surface electromyography (sEMG) signal is presented. The first step of this method is to record sEMG signal from the subjects right forearm and to acquired video frames of hand at the same time. The second step is to synchronize the beginning and the end of recorded video frame and obtain sEMG signals. The third step is to extract some most commonly used feature vectors for sEMG in the literature. Finally, feature vectors sets are fed to the ANN to estimate angle of hand movements. The obtained success rate of the ANN is given as 94.06% in the train set and 93.41% in the test set.