Emre Gürbüz

Classifying discrete interval densities of EEG signals by using DWT and SVM

This study concentrates on detection of the epileptic activities in the electroencephalogram (EEG) signals. For this aim, features are extracted from the EEG signals by using first wavelet transform and then the approach of densities based on equal frequency discretization, and these features are classified by using support vector machines. The obtained results are compared with the results of three different studies. The results show that the feature extraction method used improves the classification success rate and SVM obtains the highest classification success rate possible in faster running time.

New intelligent diagnosis method to determine thyroid disorders

In this study, the Learning Vector Quantization (LVQ) network model is enhanced with attaching conscience mechanism to be used in diagnosis of thyroid disorders, so that networks disease diagnostic success percentage is increased and the conscience coefficients which is used in the network is obtained in an adaptive manner. It is observed that the LVQ network model created with this new mechanism learns faster than the other networks, besides conscience mechanism increases percentage of networks performance. The results obtained are discussed by comparison with similar studies.

Diagnosis of diabetes by using Adaptive SVM and feature selection

In this study, a new Support Vector Machine (SVM) based method for diagnosis of diabetes is proposed. In the proposed method, feature of adaptibility is added to the support vector machine. Thus, a new kind of SVM named “Adaptive SVM” is proposed, and by using it together with the Feature Selection Method, smartly diagnosis of diseases is aimed. During the training and testing of this newly designed smart system, diabetes data set which is obtained from the medical database of University of California is used. It is observed that classification rate of this newly proposed method on the diabetes daha set is more successful than the similar studies which are implemented so far and which are in the literature.

Rotation invariant copy move forgery detection method

In recent years, due to the increment in the image editing software applications and the easiness of using these, the probability of malicious changes on the images has also increased. Copy-move forgery is one of the most widely applied modification types on the images. In case of that the copied region is rotated before being pasted, forgery detection becomes difficult. Many researchers try on proposing new rotation-invariant techniques for detecting forgeries. For this purpose, various techniques such as Zernike invariants and log-polar transform have been utilized. In this study, Circular Projection technique is used for generating feature vectors from the image blocks. When compared with the results of the studies in the same field, it is observed that the proposed method gives better results even on the rotation operations with greater angles. Experimental results show that the proposed technique is robust against scaling and mirroring operations.