Yazar "Yurtay, Nilüfer" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim
    Öğe
    Decrypting the Transposition Cipher Using a New Move Operator on Particle Swarm Optimization
    (Ieee, 2018) Demirci, Hüseyin; Yurtay, Nilüfer; Yıldız, Tuba Karagül
    Transposition cipher encryption is a historical encryption method which is based on swapping letter of columns on base text according to a secret key. In this paper, we present a new swap based move operator for the particle swarm optimization algorithm. We have tested the operator on transposition cipher encryption with a text size of 125, 250, 500 and 750 letter and 5, 10, 15 key lengths. The algorithm we developed has recovered the exact original key in some tests. In most cases, particularly, the algorithm have recovered 70% of the keys and in the worst case the algorithm have recovered only sequences of the original key which can make the encrypted text barely readable for the human eye and most of the words are decrypted.
  • Küçük Resim
    Öğe
    Diagnosing Hematological Disorders Using Deep Learning Method
    (2021) Yurtay, Nilüfer; Öneç, Birgül; Karagül, Tuba
    Deciding on the diagnosis of the disease is an important step for treating the patients. Also, the numerical value of blood tests, the personal information of patients, and most importantly, an expert opinion is necessary to diagnose a disease. With the development of technology, patient-related data are obtained both rapidly and in large sizes. Deep learning methods, which can produce meaningful results by processing the data in raw form, are beginning to give results that are close to human opinion nowadays. The present work is aimed to develop a system that will enable the diagnosis of anemia in general practice conditions due to the increasing number of patients and the intention of the hospitals, as well as the difficulties in reaching the expert medical consultant. The main contribution of this work is to make a diagnosis like a doctor with the data as the way the doctor uses it. The data set was obtained from the actual hospital environment and no intervention, such as increasing or decreasing the number of data, increasing or decreasing the number of attributes, reduction, integration, imputation, transformation, or discretization, has been made on the incoming patient data. The original hospital data are classified for the diagnosis of anemia types and the accuracy of 84,97% achieved by using a deep learning algorithm.