Scopus İndeksli Yayınlar Koleksiyonu
Permanent URI for this collectionhttps://hdl.handle.net/20.500.12416/8651
Browse
3 results
Search Results
Article Comprehensive Analysis of Data Augmentation Methods in Classification for an Imbalanced Epilepsy Dataset(Institute of Electrical and Electronics Engineers Inc., 2026) Calis, A.G.; Ergezer, H.Imbalanced class distribution reduces the generalizability of classifiers in EEG-based epilepsy detection. This study examines the impact of the synthetic minority oversampling technique (SMOTE) and its variants on imbalanced electroencephalography (EEG) data, utilizing an end-to-end data processing pipeline. Band-limited filtering is applied as pre-processing, and then the training data is gradually oversampled by 20% increments in four scenes. Experiments are conducted on coarse-k-nearest neighbor (Coarse-KNN), bagged trees, and artificial neural network (ANN) classifiers, and evaluation is performed using accuracy, precision, recall, F1 score, and Matthew’s correlation coefficient (MCC) metrics. In Scene #4, where the inter-class imbalance is eliminated, Borderline-SMOTE yielded the highest and most consistent results (F1 Score = 0.903–0.937, MCC = 0.830–0.894). Safe level-SMOTE (SL-SMOTE) and SMOTE/Geometric-SMOTE(G-SMOTE) produced second-ranked results. The findings demonstrate that appropriate variant selection provides consistent gains even across classifiers, making Borderline-SMOTE the recommended approach for imbalanced EEG classification. Furthermore, in the detailed analysis of ensemble sampling limits, SMOTE-based combined approaches (e.g., SL + G SMOTE) also produced consistent results. Basic descriptive statistics (mode, median, variance, and kurtosis) of the synthetic samples were found to be comparable to those of the real data, providing additional evidence of distributional consistency. © 2013 IEEE.Conference Object Predicting Electric Vehicle Adoption in the Eu: Analyzing Classification Performance and Influencing Attributes Across Countries, Gender, and Education Level(Institute of Electrical and Electronics Engineers Inc., 2024) Kumbasar, M.; Tokdemir, G.; Labben, T.G.; Ertek, G.Electric vehicles (EVs) have been one of the trending technologies in recent decades, as they are expected to transform the current automotive technology and transportation systems. To this end, the scope of this study is analyzing survey data on European consumers' EV purchase decisions. The objective is comparing the predictive quality of various classification algorithms in predicting EV adoption, across country, gender and education level of the participants, as well as the analysis of the influencing attributes. Initially, the data is filtered for each value of the chosen categorical attribute (country, gender or education level) with the missing values being imputed. Then, several classification algorithms in the Python sklearn package are applied through 5-fold-cross validation and the performance of the algorithms are compared based on standard classification metrics. There are notable variations in classification performance and influencing attributes depending on the values of the selected categorical attributes. © 2024 IEEE.Article A Classifier for Automatic Categorisation of Chronic Venous Insufficiency Images(Kaunas Univ Technology, 2024) Karadeniz, Talha; Tokdemir, Gul; Maras, H. Hakan; Hakan Maras, H.Chronic venous insufficiency (CVI) is a serious disease characterised by the inability of the veins to effectively return blood from the legs back to the heart. This condition represents a significant public health issue due to its prevalence and impact on quality of life. In this work, we propose a tool to help doctors effectively diagnose CVI. Our research is based on extracting Visual Geometry Group network 16 (VGG-16) features and integrating a new classifier, which exploits mean absolute deviation (MAD) statistics to classify samples. Although simple in its core, it outperforms state-of-the-art method which is known as the CVI-classifier in the literature, and additionally it performs better than the methods such as multi-layer perceptron (MLP), Naive Bayes (NB), and gradient boosting machines (GBM) in the context of VGG-based classification of CVI. We had 0.931 accuracy, 0.888 Kappa score, and 0.916 F1-score on a publicly available CVI dataset which outperforms the state-of-the-art CVI-classifier having 0.909, 0.873, and 0.900 for accuracy, Kappa score, and F1-score, respectively. Additionally, we have shown that our classifier has a generalisation capacity comparable to support vector machines (SVM), by conducting experiments on eight different datasets. In these experiments, it was observed that our classifier took the lead on metrics such as F1-score, Kappa score, and receiver operating characteristic area under the curve (ROC AUC).