Sürdürülebilir Kalkınma Amaçları
Yayınlar
- MakaleAçık Erişim2026Predicting middle school students' academic achievement in mathematics with aAnew hybrid approach
Dogan, Gurkan; Demiralp, Demet
The academic achievement of middle school students is one of the critical steps in their educational journey that affects their future academic and career prospects. During this transition period, students are faced with a variety of challenges related to their families, schools, and the individual themselves. Research shows that academic performance during these formative years is critical for future educational success and overall life success. Therefore, addressing students' academic achievement, which depends on various factors during this period, is of particular importance for both institutional and individual planning and orientation in the future. In this study, a new hybrid approach based on artificial neural networks that enables automatic analysis of data on family, school, and individual factors affecting middle school students' academic achievement in mathematics is proposed. A publicly available student performance dataset was used for training and testing the proposed hybrid approach and other models. This dataset consists of data such as mathematics grades, family information, residential information, and health status information for 395 students enrolled in two public schools in the Alentejo region of Portugal. The proposed approach achieved an R2 score of 88.6% in experimental studies with this data set, providing approximately 3% higher accuracy than its closest competitor among other methods in the literature.
AtıfOpenAlex1
İncele - MakaleYalnız künye2026Scopus SJRQ1Modeling stack voltage and remaining useful life of proton exchange membrane fuel cells using deep neural networks
Dogan, Gurkan; Imak, Andaç
Currently, various energy conversion systems are being developed to reduce environmental impacts. Proton exchange membrane fuel cells (PEMFCs), which stand out with their high efficiency characteristics, have emerged as an important alternative. However, these advantages, along with high costs and uncertainties regarding remaining useful life (RUL) estimation, restrict their further commercialization, especially in the automotive industry. This study presents a forecasting framework using a long short-term memory (LSTM) architecture enhanced with the Bahdanau attention mechanism for predicting RUL and long-term degradation trends. The model thus exhibits a performance that can more effectively capture dependencies in time series and adaptively weight significant time steps. Furthermore, the tree-structured parzen estimator (TPE) was chosen for the appropriate selection of hyperparameters in the developed forecasting framework. The validity, stability, and performance of the proposed model under different operating conditions were examined on the FC1 and FC2 datasets. Experimental studies have shown that the proposed method is the most successful among the best methods, achieving RE metric values of 1.628 × 10−3, 0.003 × 10−1, 99.16 × 10−2, and 0.044 × 10−1 respectively in estimating the best RMSE, MAPE, R2, and RUL.
- MakaleYalnız künye2026Scopus SJRQ1FedMedIDS: A non-IID-resilient federated DCNN framework for securing IoMT and IoHT devices
Gencer, Burak; Dogan, Gurkan
The diversity of devices in the Internet of Medical Things (IoMT) and Internet of Health Things (IoHT) networks creates significant security vulnerabilities against cyberattacks. Traditional centralized machine learning approaches, however, are not always suitable for these environments due to data privacy regulations. In this study, a new Deep Convolutional Neural Network (DCNN) architecture is proposed for attack detection in IoMT/IoHT networks. The proposed architecture integrates the Residual Squeeze-and-Excitation (RSE) Block, the Squeeze-and-Excitation attention mechanism (SEM), and a one-dimensional multi-pooling approach. The model was evaluated within a federated learning (FL) framework called Federated Medical Intrusion Detection System (FedMedIDS), which preserves data privacy and demonstrates robustness in heterogeneous client scenarios. Four different training strategies—Centralized, FedAvg, FedProx, and Local-Only—were adopted for the evaluations. The proposed model and existing models were tested on the Canadian Institute for Cybersecurity – Internet of Medical Things 2024 (CICIoMT2024) and Edith Cowan University – Internet of Health Things (ECU-IoHT) datasets and compared with current deep learning models. The DCNN proposed for the CICIoMT2024 dataset and a Dirichlet distribution with α = 1.0 achieved F1 scores of up to 99.92% under FedAvg and 94.64% under FedProx with α = 0.5. At α = 0.05, the highest level of heterogeneity, it outperformed the Long Short-Term Memory (LSTM) by as much as 27%. On the ECU-IoHT dataset, F1 scores of up to 89.57% under FedAvg and 89.50% under FedProx were achieved at α = 1.0. This represents an improvement of up to 37% over LSTM at the same level of heterogeneity.
- MakaleAçık Erişim2025Scopus SJRQ2Modeling and analysis of WEDM with a new deep neural network
Mahmat, Aybars; Dogan, Gurkan; Ergen, Burhan; Tosun, Nihat
In some cases, the models targeted in machining processes are difficult to apply to industrial production. Artificial intelligence techniques applied to industrial production methods in recent times eliminate these problems in machining. The use of deep learning methods in solving these problems has recently increased. In this study, a new deep neural network (DNN) based on artificial neural networks is proposed to model the machining of AISI4140 steel by wire electrical discharge machining (WEDM). The proposed DNN model predicts surface roughness (SR) and wire wear ratio (WWR), which are the most important inferences in cutting with WEDM. Obtaining inferences such as SR and WWR in a real experimental environment means a lot of time and cost. With the proposed model, there is a significant reduction in time and cost factors. The dataset used to train and test the proposed model was obtained from the real experimental environment. For training and testing the model, the data set is divided into two parts: 20% test data and 80% training data. In the experimental tests carried out with the proposed model, error rates such as Mean Absolute Error (MAE), R2, and Mean Squared Error (MSE) accuracy rate between SR and WWR values estimated according to the SR and WWR values obtained from the real environment were examined. MAE, MSE, and R2 for SR reached 0.0495, 0.0005, and 98.48%, respectively, while MAE, MSE, and R2 for WWR reached 0.0156, 0.0036, and 98.39%, respectively. These results of the proposed model provided a lower error rate and higher accuracy than the state-of-the-art models in the literature.
AtıfScopus5WoS5OpenAlex7
İncele - BildiriYalnız künye2024Performance Analysis of Attention Mechanisms for Melanoma Cancer Detection
Dogan, Gurkan
The deadliest type of skin cancer worldwide is melanoma. If melanoma skin cancer is detected and treated at an early stage, the mortality rate can be significantly reduced. In particular, convolutional neural networks, one of the deep learning techniques that have achieved remarkable success in recent years in areas such as computer vision, natural language processing, signal processing, etc., can detect this type of cancer automatically, quickly and with high accuracy. Moreover, the accuracy performance of the neural network can be further increased by using attention mechanisms together with convolutional neural networks. However, attention mechanisms have not been used correctly and effectively in the neural network models proposed so far for the detection of melanoma skin cancer. This study examines the effect of attention mechanisms on the accuracy performance of a convolutional neural network and guides researchers to use these mechanisms correctly and effectively. To achieve this, a base model based on a convolutional neural network was built. Then, seven different attention mechanisms were integrated separately into this basic model and the results of the experimental tests of each were presented.
AtıfScopus1OpenAlex1İncele