Arif Gülten

Classifier ensemble construction with rotation forest to improve medical diagnosis performance of machine learning algorithms

Improving accuracies of machine learning algorithms is vital in designing high performance computer-aided diagnosis (CADx) systems. Researches have shown that a base classifier performance might be enhanced by ensemble classification strategies. In this study, we construct rotation forest (RF) ensemble classifiers of 30 machine learning algorithms to evaluate their classification performances using Parkinsons, diabetes and heart diseases from literature. While making experiments, first the feature dimension of three datasets is reduced using correlation based feature selection (CFS) algorithm. Second, classification performances of 30 machine learning algorithms are calculated for three datasets. Third, 30 classifier ensembles are constructed based on RF algorithm to assess performances of respective classifiers with the same disease data. All the experiments are carried out with leave-one-out validation strategy and the performances of the 60 algorithms are evaluated using three metrics; classification accuracy (ACC), kappa error (KE) and area under the receiver operating characteristic (ROC) curve (AUC). Base classifiers succeeded 72.15%, 77.52% and 84.43% average accuracies for diabetes, heart and Parkinsons datasets, respectively. As for RF classifier ensembles, they produced average accuracies of 74.47%, 80.49% and 87.13% for respective diseases. RF, a newly proposed classifier ensemble algorithm, might be used to improve accuracy of miscellaneous machine learning algorithms to design advanced CADx systems.

Genetic algorithm wrapped Bayesian network feature selection applied to differential diagnosis of erythemato-squamous diseases

This paper presents a new method for differential diagnosis of erythemato-squamous diseases based on Genetic Algorithm (GA) wrapped Bayesian Network (BN) Feature Selection (FS). With this aim, a GA based FS algorithm combined in parallel with a BN classifier is proposed. Basically, erythemato-squamous dataset contains six dermatological diseases defined with 34 features. In GA-BN algorithm, GA makes a heuristic search to find most relevant feature model that increase accuracy of BN algorithm with the use of a 10-fold cross-validation strategy. The subsets of features are sequentially used to identify six dermatological diseases via a BN fitting the corresponding data. The algorithm, in this case, produces 99.20% classification accuracy in the diagnosis of erythemato-squamous diseases. The strength of feature model generated for BN is furthermore tested with the use of Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), Simple Logistics (SL) and Functional Decision Tree (FT). The resultant classification accuracies of algorithms are 98.36%, 97.00%, 98.36% and 97.81% respectively. On the other hand, BN algorithm with classification accuracy of 99.20% is quite a high diagnosis performance for erythemato-squamous diseases. The proposed algorithm makes no more than 3 misclassifications out of 366 instances. Furthermore, FS power of GA is also compared with two alternative search algorithms, i.e. Best First (BF) and Sequential Floating (SF). The obtained results have all together shown that the proposed GA-BN based FS and prediction strategy is very promising in diagnosis of erythemato-squamous diseases.

A robust color image watermarking with Singular Value Decomposition method

The performance of a watermarking method based on Singular Value Decomposition (SVD) has been improved for color image in this paper. One of the common methods used for hiding information on image files is Singular Value Decomposition method which used in the frequency domain. In Singular Value Decomposition based watermarking techniques; watermark embedding can usually be achieved by modifying the least significant bits of the singular value matrix. This paper gives application results which show the watermarking security of using this algorithm for the watermarking and demonstrate the accuracy of these methods. The performance comparison of the algorithms was also realized.

A Robust Multi-Class Feature Selection Strategy Based on Rotation Forest Ensemble Algorithm for Diagnosis of Erythemato-Squamous Diseases

In biomedical studies, accuracy of classification algorithms used in disease diagnosis systems is certainly an important task and the accuracy of system is strictly related to extraction of discriminatory features from data. In this paper, we propose a new multi-class feature selection method based on Rotation Forest meta-learner algorithm. The feature selection performance of this newly proposed ensemble approach is tested on Erythemato-Squamous diseases dataset. The discrimination ability of selected features is evaluated by the use of several machine learning algorithms. In order to evaluate the performance of Rotation Forest Ensemble Feature Selection approach quantitatively, we also used various and widely utilized ensemble algorithms to compare effectiveness of resultant features. The new multi-class or ensemble feature selection algorithm exhibited promising results in eliminating redundant attributes. The Rotation Forest selection based features demonstrated accuracies between 98% and 99% in various classifiers and this is a quite high performance for Erythemato-Squamous Diseases diagnosis.In biomedical studies, accuracy of classification algorithms used in disease diagnosis systems is certainly an important task and the accuracy of system is strictly related to extraction of discriminatory features from data. In this paper, we propose a new multi-class feature selection method based on Rotation Forest meta-learner algorithm. The feature selection performance of this newly proposed ensemble approach is tested on Erythemato-Squamous diseases dataset. The discrimination ability of selected features is evaluated by the use of several machine learning algorithms. In order to evaluate the performance of Rotation Forest Ensemble Feature Selection approach quantitatively, we also used various and widely utilized ensemble algorithms to compare effectiveness of resultant features. The new multi-class or ensemble feature selection algorithm exhibited promising results in eliminating redundant attributes. The Rotation Forest selection based features demonstrated accuracies between 98% and 99% in various classifiers and this is a quite high performance for Erythemato-Squamous Diseases diagnosis.

Modelling and simulation of the multi-scroll chaotic attractors using bond graph technique

Abstract This paper presents modelling and simulation of multi-scroll chaotic attractors by using a new simple and more general bond graph model. For this purpose, the multi-segment non-linear resistor in Chua’s circuit is modelled by using piecewise linearization with control inequalities. The proposed model consists of active/passive circuit elements, voltage-controlled current source (VCCS) and ideal switches. The advantage of modelling multi-segment non-linear resistor by using control inequalities yields minimum number of the switches and sources. Proposed model is simple and more general and, especially, could be used in various kinds of non-linear circuit in the chaos studies. Generally, two different non-linear resistor models are used in the literature to obtain odd and even numbers of the scrolls. In this study, one model is developed for both multi-scroll chaotic attractors. In this paper, bond graph simulation of Chua’s circuit is realized by using proposed model. The BONDAS program that developed in Matlab is used for the simulations, and satisfactory results are obtained.

A New Watermarking System Based on Discrete Cosine Transform (DCT) in Color Biometric Images

This paper recommend a biometric color images hiding approach An Watermarking System based on Discrete Cosine Transform (DCT), which is used to protect the security and integrity of transmitted biometric color images. Watermarking is a very important hiding information (audio, video, color image, gray image) technique. It is commonly used on digital objects together with the developing technology in the last few years. One of the common methods used for hiding information on image files is DCT method which used in the frequency domain. In this study, DCT methods in order to embed watermark data into face images, without corrupting their features.

Analysis of switched systems using the bond graph methods

Abstract In recent years, the analysis of switching systems has gained importance. In this paper, the formulation of state and output equations and solutions of switched-systems are presented by using the bond graph model with a new simple and more general switch definition. The theory is illustrated by a few examples and the output of the computer programme called BOMAS is presented.

Swarm optimized organizing map (SWOM): A swarm intelligence basedoptimization of self-organizing map

This work studies the optimization of SOM algorithm in terms of reducing its training time by the use of a swarm intelligence method, i.e. particle swarm optimization (PSO). Our novel algorithm optimizes SOM with PSO and reduces computational time of the training phase of SOM significantly. The performance of the algorithms has been tested with genomic datasets, biomedical datasets and an artificial dataset to show the efficiency of swarm optimized SOM, i.e. SWOM. The experimental comparison between SOM and SWOM, has demonstrated significant reduction in training time of SWOM with preservation of clustering quality.

Assessing Effects of Pre-Processing Mass Spectrometry Data on Classification Performance:

Disease prediction through mass spectrometry (MS) data is gaining importance in medical diagnosis. Particularly in cancerous diseases, early prediction is one of the most life saving stages. High dimension and the noisy nature of MS data requires a two-phase study for successful disease prediction; first, MS data must be pre-processed with stages such as baseline correction, normalizing, de-noising and peak detection. Second, a dimension reduction based classifier design is the main objective. Having the data pre-processed, the prediction accuracy of the classifier algorithm becomes the most significant factor in the medical diagnosis phase. As health is the main concern, the accuracy of the classifier is clearly very important. In this study, the effects of the pre-processing stages of MS data on classifier performances are addressed. Three pre-processing stages—baseline correction, normalization and de-noising—are applied to three MS data samples, namely, high-resolution ovarian cancer, low-resolution prostate cancer and a low-resolution ovarian cancer. To measure the effects of the pre-processing stages quantitatively, four diverse classifiers, genetic algorithm wrapped K-nearest neighbor (GA-KNN), principal component analysis-based least discriminant analysis (PCA-LDA), a neural network (NN) and a support vector machine (SVM) are applied to the data sets. Calculated classifier performances have demonstrated the effects of pre-processing stages quantitatively and the importance of pre-processing stages on the prediction accuracy of classifiers. Results of computations have been shown clearly.

Examination of chaotic behaviours using bond graph model

In this paper, modelling and simulation of Chuas chaotic oscillator, which exhibits rich chaotic behaviours, are presented by using the bond graph model. Up to now modelling of Chuas chaotic oscillator using bond graph model is not yet developed. The non-linear resistor in the circuit is modelled in this contribution by linear time-invariant components and ideal switches using piecewise linearization approach. The bond graph model of all the circuit including switches is then generated. Simulations are provided via the computer program called as BOMAS using the obtained bond graph model. Finally, Chuas circuit is verified experimentally. It is shown that all experimental and simulation results well agree with the chaotic behaviours of Chuas circuit.