Yazar "Budak, Cafer" seçeneğine göre listele
Listeleniyor 1 - 20 / 21
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe An atom search optimization approach for IIR system identification(Taylor & Francis Inc, 2023) Ekinci, Serdar; Budak, Cafer; Izci, Davut; Gider, VeyselFiltering, or digital signal processing, is a significant and fundamental requirement in fields such as signal systems and computers. The process of designing optimal digital filters is difficult, which has led researchers to design filters using emerging evolutionary computations. Metaheuristics have emerged as the most promising tool for solving optimization problems, with excellent development and improvement. However, it has not been clear how to select the best performing metaheuristic to design an optimal digital filter. In this paper, a digital infinite impulse response (IIR) filter is constructed using the atom search optimization (ASO) algorithm impressed by the physical motion of atoms in nature based on molecular dynamics. The simulation results obtained are extensively compared with the results of other optimization algorithms such as moth flame optimization, gravitational search algorithm and artificial bee colony optimization. ASO was found to have the highest percentage of improvement. Furthermore, eight cases are analyzed across four numerical filter instances with the same degree and four with reduced degree, and the results are validated by outperforming several different algorithm-based approaches in the literature. The stability analysis on the basis of pole zero diagrams further cements the efficacy of the ASO for IIR system identification problem.Öğe Automatic cell nuclei segmentation using superpixel and clustering methods in histopathological images(Balkan Yayın, 2021) Mendi, Gamze; Budak, CaferIt is seen that there is an increase in cancer and cancer-related deaths day by day. Early diagnosis is vital for the early treatment of the cancerous area. Computer-aided programs allow for the early diagnosis of unhealthy cells that specialist pathologists diagnose due to efforts. In this study, clustering and superpixel segmentation techniques were used to detect cell nuclei in high-resolution histopathology images automatically. As a result of the study, the successful performances of the segmentation algorithms were analyzed and evaluated. It is seen that better success is obtained in the Watershed and FCM algorithms in highresolution histopathological images used. Quickshift and SLIC methods gave better results in terms of precision. It is seen that there are k-Means and FCM algorithms that provide the best performance in F measure (F-M), and the correct negative rate (TNR) is more successful in Quickshift, kMeans, and SLIC methods.Öğe Biomedical image segmentation with modified U-Net(International Information and Engineering Technology Association, 2023) Tatlı, Umut; Budak, CaferImage segmentation is an important field in image processing and computer vision, particularly in the development of methods to assist experts in the biomedical and medical fields. It plays a vital role in saving time and costs. One of the mostsuccessful and significant methods in image segmentation using deep learning is the U-Net model. In this paper, we propose U-Net11, a novel variant of U-Net that uses 11 convolutional layers and introduces some modifications to improve the segmentation performance. The classical U-Net model was developed and tested on three different datasets, outperforming the traditional U-Net approach. The U-Net11 model was evaluated for breast cancer segmentation, lung segmentation from CT images, and the nuclei segmentation dataset from the Data Science Bowl 2018 competition. These datasets are valuable due to their varying image quantities and the varying difficulty levels in segmentation tasks. The modified U-Net model has achieved Dice Similarity Coefficient scores of 69.09% on the breast cancer dataset, 95.02% on the lung segmentation dataset and 81.10% on the nuclei segmentation dataset, exceeding the performance of the classical U-Net model by 5%, 2% and 4% respectively. This difference in success rates is particularly significant for critical segmentation datasets.Öğe Classification of the Ionospheric Disturbances Caused by Geomagnetic and Seismic Activity with K-Nearest Neighbors Algorithm(Springer, 2024) Budak, Cafer; Karatay, Secil; Erken, Faruk; Cinar, AliDetection of earthquake-precursor signals a few days before the earthquake day has become an area of increasing interest. In recent years, it has been observed that the major earthquakes and geomagnetic activity can cause significant disturbances and anomalies in the ionospheric parameters such as Total Electron Content (TEC). TEC provides important information about the detection of anomalies and disturbances related to seismic and geomagnetic activity in the ionosphere. The main goal of this study is to classify the disturbances due to the seismic and geomagnetic activity in the ionosphere using TEC data. For this purpose, the K-Nearest Neighbors (K-NN) algorithm is applied to TEC estimated from Global Positioning System stations during five earthquakes with magnitudes Mw greater than 5.6 between 1999 and 2016 and for the geomagnetically quiet and disturbed conditions of the ionosphere. The data is divided into four classes as non-earthquake and non-geomagnetic activity, geomagnetic activity, possible earthquake precursor and earthquake for each earthquake. The study is performed into two groups as Group I, where five days before the earthquake day are marked as precursor, and Group II, where three days are marked as precursors. It is observed that 5055 test samples out of a total of 5184 are classified as true whereas 129 are classified as false for Group I. 3912 are classified as true and 105 are classified as false for Group II. For the possible earthquake class, the Accuracy values increase inversely with the distance of the stations from the epicenter and directly related to the magnitude of the earthquakes.Öğe Daily solar radiation prediction using LSTM neural networks(Institute of Electrical and Electronics Engineers Inc., 2022) Gider, Veysel; Budak, Cafer; İzci, Davut; Ekinci, SerdarThe integration of solar energy with the smart grids and existing infrastructure makes it a cost-effective and environmentally-friendly solution to address the growing energy need. To make use of the potential of solar energy, several challenges such as the stability of generated energy and the supply-demand imbalance must be overcome. In this regard, an accurate forecast model for global solar radiation (GSR) can be useful for power generation planning and system reliability. The GSR estimate is regarded as the most significant and critical element in defining solar system characteristics, thus, it is crucial in predicting the generated energy. This work, therefore, employs long-short-term memory (LSTM) as a deep learning method to successfully estimate solar irradiance and capture the stochastic fluctuations. In this respect, the measurement data (from year 2021) obtained from the station installed in Dicle University (Turkey), Science and Technology Application and Research Centre (DUBTAM) were used, and the efficiency of the proposed method was evaluated. © 2022 IEEE.Öğe Detection of object (Weapons) with deep learning algorithms from images obtained by unmanned aerial vehicles(Dicle Üniversitesi Mühendislik Fakültesi, 2022) Burgaz, Mustafa; Budak, CaferToday, the rapid development of Artificial Intelligence technologies is effective in the success of deep learning algorithms in different application areas. These applications detect many objects that even the human eye cannot detect in object detection in videos and images with deep learning algorithms.In this study, it is aimed to detect weapons from images obtained from Unmanned Aerial Vehicle (UAV) by using deep learning algorithms. Images were obtained from the UAV at 200 different angles and heights. Images from different angles and heights obtained from the unmanned aerial vehicle are trained by Regional Based Convolutional Neural Networks (R-CNN) and Residual Neural Network (ResNet). Twothirds of the images we obtained were split into training images and one-third into test images. The feature maps extracted from the images used for training were compared with the test images. By bringing these compared images closer to the desired images, 99% of the desired image detection is achieved. Performance evaluation of the algorithms was made using Loss plot, mAP curves, Precision, Recall and F1-Score. The performance evaluation of the detected images is discussed, and the success of deep learning algorithms used in object detection is presented. The ResNet model showed higher performance with 64% accuracy, 94% recall and 76% F1 score.Öğe Detection of ring cell cancer in histopathological images with region of interest determined by SLIC superpixels method(Springer, 2022) Budak, Cafer; Mençik, VasfiyeGastric cancer is the sixth most common cancer and the fourth leading cause of cancer deaths worldwide. Gastric cancer presents with a more insidious onset and is most frequently discovered at an advanced stage. Early diagnosis is critical since the stage of the disease is determinant in the severity, treatment, and survival rate of cancer. In the study, the Region of Interest (RoI) was determined in histopathological images using image preprocessing techniques and signet ring cell carcinoma (SRCC) was detected with popular deep learning models VGG16, VGG19, and InceptionV3. The fine-tuning strategy was applied by customizing the last five layers of deep network models based on the target data. The parameters of accuracy, precision, recall, and F1-score were used to evaluate the model performance. Signet ring cell dataset taken from the competition "Digestive System Pathological Detection, and Segmentation Challenge 2019" was employed. When compared to results of the DigestPath2019 Grand challenge ring cell gastric cancer competition, higher accuracy rates were obtained using deep learning models with the accurate defined RoI images. VGG16 model exhibited a higher performance with accuracy of 95% and a F1-score of 95% among the models. The results obtained by the algorithm were analyzed and confirmed by the experienced pathologist.Öğe Determining similarities of COVID-19-lung cancer drugs and affinity binding mode analysis by graph neural network-based GEFA method(Taylor & Francis, 2021) Budak, Cafer; Mençik, Vasfiye; Gider, VeyselCOVID-19 is a worldwide health crisis seriously endangering the arsenal of antiviral and antibiotic drugs. It is urgent to find an effective antiviral drug against pandemic caused by the severe acute respiratory syndrome (Sars-Cov-2), which increases global health concerns. As it can be expensive and time-consuming to develop specific antiviral drugs, reuse of FDA-approved drugs that provide an opportunity to rapidly distribute effective therapeutics can allow to provide treatments with known preclinical, pharmacokinetic, pharmacodynamic and toxicity profiles that can quickly enter in clinical trials. In this study, using the structural information of molecules and proteins, a list of repurposed drug candidates was prepared again with the graph neural network-based GEFA model. The data set from the public databases DrugBank and PubChem were used for analysis. Using the Tanimoto/jaccard similarity analysis, a list of similar drugs was prepared by comparing the drugs used in the treatment of COVID-19 with the drugs used in the treatment of other diseases. The resultant drugs were compared with the drugs used in lung cancer and repurposed drugs were obtained again by calculating the binding strength between a drug and a target. The kinase inhibitors (erlotinib, lapatinib, vandetanib, pazopanib, cediranib, dasatinib, linifanib and tozasertib) obtained from the study can be used as an alternative for the treatment of COVID-19, as a combination of blocking agents (gefitinib, osimertinib, fedratinib, baricitinib, imatinib, sunitinib and ponatinib) such as ABL2, ABL1, EGFR, AAK1, FLT3 and JAK1, or antiviral therapies (ribavirin, ritonavir-lopinavir and remdesivir).Öğe Drug Solubility Prediction: A Comparative Analysis of GNN, MLP, and Traditional Machine Learning Algorithms(Gazi University, 2024) Gider, Veysel; Budak, CaferThe effective development and design of pharmaceuticals hold fundamental importance in the fields of medicine and the pharmaceutical industry. In this process, the accurate prediction of drug molecule solubility is a critical factor influencing the bioavailability, pharmacokinetics, and toxicity of drugs. Traditionally, mathematical equations based on chemical and physical properties have been used for drug solubility prediction. However, in recent years, with the advancement of artificial intelligence and machine learning techniques, new approaches have been developed in this field. This study evaluated different modeling approaches consisting of Graph Neural Networks (GNN), Multilayer Perceptron (MLP), and traditional Machine Learning (ML) algorithms. The Random Forest (RF) model stands out as the optimal performer, manifesting superior efficacy through the attainment of minimal error rates. It attains a Root Mean Square Error (RMSE) value of 1.2145, a Mean Absolute Error (MAE) value of 0.9221, and an R-squared (R2) value of 0.6575. In contrast, GNN model displays comparatively suboptimal performance, as evidenced by an RMSE value of 1.8389, an MAE value of 1.4684, and an R2 value of 0.2147. These values suggest that the predictions of this model contain higher errors compared to other models, and its explanatory power is lower. These findings highlight the performance differences among different modeling approaches in drug solubility prediction. The RF model is shown to be more effective than other methods, while the GNN model performs less effectively. This information provides valuable insights into which model should be preferred in pharmaceutical design and development processes.Öğe Effect on model performance of regularization methods(Dicle Üniversitesi Mühendislik Fakültesi, 2021) Budak, Cafer; Mençik, Vasfiye; Asker, Mehmet EminArtificial Neural Networks with numerous parameters are tremendously powerful machine learning systems. Nonetheless, overfitting is a crucial problem in such networks. Maximizing the model accuracy and minimizing the amount of loss is significant in reducing in-class differences and maintaining sensitivity to these differences. In this study, the effects of overfitting for different model architectures with the Wine dataset were investigated by Dropout, AlfaDropout, GausianDropout, Batch normalization, Layer normalization, Activity normalization, L1 and L2 regularization methods and the change in loss function the combination with these methods. Combinations that performed well were examined on different datasets using the same model. The binary cross-entropy loss function was used as a performance measurement metric. According to the results, the Layer and Activity regularization combination showed better training and testing performance compared to other combinations.Öğe Elektromanyetik dalgaların biyolojik dokular içindeki yayılımınının zamanda sonlu farklar metodu ile analizi ve simülasyonu(2017) Budak, Cafer; Kurt, M. BahattinBu çalışmada, güçlü sayısal yöntemlerden biri olan FDTD (Zamanda Sonlu Farklar Yöntemi) ayrıntılı olarak incelenmiştir. FDTD yönteminin hızlı bilgisayarlarda kullanılması, hafızanın artması hesaplama performansının gelişmesi FDTD kullanarak dalga simülasyonunu mümkün kıldı. Başlangıçta FDTD ile ilgili matematiksel ve fiziksel özellikler verilerek konu ile ilgili teorik bilgi eksikliği giderilmiş. Ardından sayısal analiz için gerekli formülasyonlar oluşturulmuş. FDTD yöntemiyle ilgili sınır koşulları, parametre seçimi, kararlılık kriterleri hakkında bilgi verilmiştir. Uygulama olarak da, Elektromanyetik dalga yayılım problemleri için geliştirilen, matlab programları ile farklı ortamlarda, farklı dalga tipleri ile dalga hareketlerinin analizleri yapılmış ve bunlarla ilgili bir iki ve üç boyutlu simülasyonlar gerçekleştirilmiştir. Son olarak da küp şeklinde kafa eşdeğeri dielektrik madde üzerindeki emilen EM enerji yani SAR grafikleri çizdirilmiştir.Öğe Instruction of molecular structure similarity and scaffolds of drugs under investigation in ebola virus treatment by atom-pair and graph network: A combination of favipiravir and molnupiravir(Elsevier Ltd, 2022) Gider, Veysel; Budak, CaferThe virus that causes Ebola is fatal. Although many researchers have attempted to contain this deadly infection, the fatality rate remains high. The atom-pair fingerprint technique was used to compare drugs suggested for the treatment of Ebola or those that are currently being tested in clinical settings. Subsequently, using scaffold network graph (SNG) methods, the molecular and structural scaffolds of the drugs chosen based on these similar results were created, and the drug structures were examined. Public databases (PubChem and DrugBank) and literature regarding Ebola treatment were used in the analysis. Graphical representations of the molecular architecture and core structures of the drugs with the highest similarity to Food and Drug Administration (FDA)-approved drugs were produced using the SNG method. The combination of molnupiravir, the first licensed oral medication candidate for COVID-19, and favipiravir, employed in other viral outbreaks, should be further researched for treating Ebola, as observed in our study. We also believe that chemists will benefit from understanding the core structure(s) of medication molecules effective against the Ebola virus, their inhibitors, and the chemical structure similarities of existing pharmaceuticals utilized to build alternative drugs or drug combinations.Öğe Investigation of electromagnetic energy accumulated in the brain by mobile phones for different frequencies(2010) Budak, Cafer; Kurt, Muhammed BahaddinThe beginning of the knowledge about the effects of electromagnetic fields on human body is almost simultaneous with the knowledge on electromagnetic fields themselves. The desire for the knowledge in order to become well aware of electromagnetic fields is increasing as electronics are spread. Especially the field of the effects of mobile phones on human brain is actually one of the most attractive topics since there has been no device which is used as near as mobile phone to the brain. The most important issue is that the heat release which caused by mobile phone. The heat release rate was defined here as Specific Absorption Rate (SAR). In this study, a model was developed by using different electrical frequencies of the brain tissues. By using this model, the SAR variation graphs for the effects of different mobile phones on brain were obtained through Finite difference Time Domain (FDTD) method.Öğe İşe alım süreçlerinde aşamalı olarak TOPSIS ve VIKOR yöntemleri uygulanarak iş gören seçimi yapılması(Dicle Üniversitesi Mühendislik Fakültesi, 2023) Aslan, Abdulbari; Hüseyinoğlu, Mesut; Budak, CaferDeğişen ve gelişen dünyada kurumların daha rekabetçi ve sürdürülebilir yönetim süreçlerinde karar verme oldukça önemli bir yer edinmiştir. Son 80 yılda tamamen sistemli bir disiplin ve model halini alan karar verme bilimi tüm organizasyon ve yaşam alanlarına yön vermeye devam etmektedir. Tüm şirketler için hayati bir öneme sahip olan ve gittikçe daha da önem kazanan çalışan etmeni ve çalışan seçimi de karar verme modelleri ile yönetilmeye başlanmıştır. Bu çalışmada öncelikle karar verme modelleri hakkında bilgiler sunulmuştur. Daha sonra Çok Kriterli Karar Verme (ÇKKV) modellerinden TOPSIS yöntemi kullanılarak iş başvurusu yapan 8 adaydan 4 kişilik mülakat listesi oluşturulmuştur. Son olarak seçilen adaylar içinden en uygun adayı tercih etmek için VIKOR yöntemi uygulanmış ve başarılı sonuçlar elde edilmiştir.Öğe LSTM based forecasting of the next day’s values of ionospheric total electron content (TEC) as an earthquake precursor signal(Springer Science and Business Media Deutschland GmbH, 2023) Budak, Cafer; Gider, VeyselThe sudden vibrations that occur due to the fractures in the Earth’s crust, spreading in waves and shaking the Earth’s surface, are natural disaster that causes significant loss of life and property. To take measures against these destructive effects, it is important to be able to forecast the occurrence time of an earthquake in advance. However, although earthquake experts can forecast which fault line the next earthquake may occur on by monitoring the movements in the fault lines, they cannot accurately forecast the exact timing. Detection of earthquake precursor signals a few days before the earthquake has become an increasingly popular field of interest. Strong correlations have been observed between earthquakes and ionospheric parameters in recent years. Total Electron Content (TEC) is an important parameter that can be affected by seismic activity in ionospheric studies and has been investigated as a potential earthquake precursor by many researchers. It has been observed that earthquakes cause significant disturbances and changes in TEC values, which are one of the ionospheric parameters. The ability to identify earthquake precursor signals before an earthquake occurs is critically important for earthquake detection. We evaluate the performance of the proposed approach using GPS-TEC data obtained from numerous ground-based GPS stations in earthquake-prone regions of Turkey, Italy, Japan, and China. In this study, TEC values in six different regions where earthquakes with Mw > 5.6 occurred were forecasted one day before the earthquake using LSTM. The results showed that the LSTM model achieved an R-square (R2) value of at least 0.9982 and the root mean square error (RMSE) value of at most 0.2302 for all experimental earthquake days used. The proposed approach may be useful for monitoring ionospheric anomalies and potentially developing an early warning system for earthquakes.Öğe Online diagnosis of COVID-19 from chest radiography images by using deep learning algorithms(Springer Science and Business Media, 2023) Budak, Cafer; Mençik, Vasfiye; Varışlı, OsmanThe COVID-19 outbreak, which has a devastating impact on the health and well-being of the global population, is a respiratory disease. It is vital to determine, isolate and treat people with the disease as soon as possible to fight against the COVID-19 pandemic. Even though the reverse transcription polymerase chain reaction (RT-PCR) test, the accuracy of which is about 63%, seems to be a good option for determining COVID-19, it is a disadvantage is that test kits are few, are difficult to obtain in remote rural areas and have low accuracy. Chest X-ray (CXR) has become essential for rapidly diagnosing the rapidly spreading COVID-19 disease worldwide, so it is urgent to develop an online system that will help specialists identify infected patients with CXR images. In this study developed a transfer learning-based diagnosis system for online diagnosis of COVID-19 patients using CXR images. Transfer learning-based deep learning models VGG16, VGG19, ResNet50, InceptionV3, Xception, MobileNet, DenseNet121 and DenseNet201 were used for the experimental studies. We explored the COVID-19 radiography database from Kaggle, which is open to the public, using image preprocessing techniques and data augmentation. The images captured by the various terminals are transferred to the web server in the created system. Similar to the ensemble learning approach, the percentage accuracy of the model with the highest prediction value among the eight deep learning models is displayed on the screen. The results show that the proposed online diagnosis system performs better than others with the highest accuracy, precision, recall and F1 values of 98%, 99%, 97% and 97%, respectively. The results show that deep learning models help to increase the efficiency of chest radiograph scanning and have promising potential in predicting COVID-19 cases. The online diagnostic system will be a helpful tool for radiologists as it diagnoses COVID-19 quickly and with high accuracy.Öğe PID controller design for DFIG-based wind turbine via reptile search algorithm(Institute of Electrical and Electronics Engineers Inc., 2022) İzci, Davut; Ekinci, Serdar; Budak, Cafer; Gider, Veysel; 0000-0001-8359-0875This paper presents a new design procedure for a doubly fed induction generator (DFIG) based wind energy conversion (WEC) system in a wind turbine (WT) using a proportional-integral-derivative (PID) controller and a recent metaheuristic approach known as reptile search algorithm (RSA). As the control scheme has a significant role on the efficiency and reliability of DFIG-based WEC system, we aim to propose the RSA tuned PID controller as the most efficient approach to operate this system. To demonstrate the efficiency and reliability of the proposed design method, previously reported design schemes such as gravitational search algorithm, bacterial foraging optimization and particle swarm optimization based PID controller approaches were used for comparisons. The obtained results showed that the proposed reptile search algorithm tuned PID controller with 6th order transfer function model of doubly fed induction generator enhances the transient performance considerably compared to other reported design approaches for wind energy conversion system. © 2022 IEEE.Öğe Recommending Tolerance Value for SpO2 Devices with Linear Regression Based on Measuring Tape(Fırat University, 2025) Üstüner, Özge; Budak, CaferMonitoring blood oxygen levels is vital for tracking various respiratory diseases like bronchitis, pneumonia, COPD, and critical care patients, including those with COVID-19. SpO2 devices, calculating oxygen percentage via finger or earlobe tissue, play a crucial role. Field studies have revealed concerns regarding the accuracy of SpO2 measurements due to high deviations, particularly in patients with low oxygen saturation, prompting the initiation of this study to ensure accurate interpretation of the device's measurements. Using a linear regression algorithm, SpO2 values from different bands were classified for quality. Tolerance values and deviation thresholds for each band were recommended. Additionally, linear regression aimed to save time by making result estimations with less data, facilitating more device monitoring and frequent testing. Results closely matched actual values, suggesting contributions to more frequent application and rapid interpretation. Following the European standard 80601-2-61, measurements were taken from three pulse oximeter brands (Contec MS100 model simulator) in three bands: 70-79%, 80-89%, and 90-100% SpO2. For each measurement, an Arms curve graph was generated using linear regression, and the mean square error (MSE) and Arms values were calculated to evaluate devices. In conclusion, deviation rates increase at low oxygen saturation levels, and recommended % SpO2 deviation values were proposed for each band and device quality.Öğe Reduction in impulse noise in digital images through a new adaptive artificial neural network model(Springer, 2015) Budak, Cafer; Turk, Mustafa; Toprak, AbdullahIn this paper, an adaptive artificial neural network model is developed in order to restore severely corrupted images. The proposed new and effective impulse noise reduction filter is named as adaptive neural network models with an algorithm based on artificial neural networks. Networks trained at different noise intensities get activated according to the intensity of the noise and estimate the most suitable neighboring pixel that can replace the corrupted pixel. The proposed algorithm reduces impulse noise effectively while also protecting the details. Experimental results show that the proposed algorithm performs better compared with other traditional filters.Öğe Removal of impulse noise in digital images with naive Bayes classifier method(Tubitak Scientific & Technological Research Council Turkey, 2016) Budak, Cafer; Turk, Mustafa; Toprak, AbdullahA new method has been presented in this paper to remove randomly formed impulse noise in digital images. This method is one of the favorite learning approaches of the Bayes learning method and is frequently called the na ve Bayes classifier. It has especially been used more frequently in recent times in the field of signal processing. Prior to restoration of the noisy pixels of the image as is done here, the image is first separated into pieces, and then an associated learning set is formed for each piece using the noise-free pixels. These learning sets that are different for each piece are used in order to estimate the pixel that will replace the noisy one. The proposed method is both simple and easy to apply. Our comprehensive experimental studies show that our proposed method outperforms other filters that are very popular in the literature.