Browsing by Author "Kina, Erol"

Now showing 1 - 6 of 6

Data Oversampling and Imbalanced Datasets: an Investigation of Performance for Machine Learning and Feature Engineering
(Springernature, 2024) Mujahid, Muhammad; Kina, Erol; Rustam, Furqan; Villar, Monica Gracia; Alvarado, Eduardo Silva; Diez, Isabel De La Torre; Ashraf, Imran
The classification of imbalanced datasets is a prominent task in text mining and machine learning. The number of samples in each class is not uniformly distributed; one class contains a large number of samples while the other has a small number. Overfitting of the model occurs as a result of imbalanced datasets, resulting in poor performance. In this study, we compare different oversampling techniques like synthetic minority oversampling technique (SMOTE), support vector machine SMOTE (SVM-SMOTE), Border-line SMOTE, K-means SMOTE, and adaptive synthetic (ADASYN) oversampling to address the issue of imbalanced datasets and enhance the performance of machine learning models. Preprocessing significantly enhances the quality of input data by reducing noise, redundant data, and unnecessary data. This enables the machines to identify crucial patterns that facilitate the extraction of significant and pertinent information from the preprocessed data. This study preprocesses the data using various top-level preprocessing steps. Furthermore, two imbalanced Twitter datasets are used to compare the performance of oversampling techniques with six machine learning models including random forest (RF), SVM, K-nearest neighbor (KNN), AdaBoost (ADA), logistic regression (LR), and decision tree (DT). In addition, the bag of words (BoW) and term frequency and inverse document frequency (TF-IDF) features extraction approaches are used to extract features from the tweets. The experiments indicate that SMOTE and ADASYN perform much better than other techniques thus providing higher accuracy. Additionally, overall results show that SVM with 'linear' kernel tends to attain the highest accuracy and recall score of 99.67% and 1.00% on ADASYN oversampled datasets and 99.57% accuracy on SMOTE oversampled dataset with TF-IDF features. The SVM model using 10-fold cross-validation experiments achieved 97.40 mean accuracy with a 0.008 standard deviation. Our approach achieved 2.62% greater accuracy as compared to other current methods.
Investigation of Effect of Exercise on Physical Parameters on Football Referees at Van Region
(Wiley-blackwell, 2015) Kina, Erol; Arihan, Okan; Erkec, Ozlem Ergul; Kara, Mehmet; Arihan, Seda Karaoz; Bektas, Yener
Low-Calorie Cookies Enhanced With Fish Oil-Based Nano-Ingredients for Health-Conscious Consumers
(Amer Chemical Soc, 2024) Meral, Raciye; Kina, Erol; Ceylan, Zafer
This study explored the effectiveness of fish oil (FO)-loaded nanoemulsions, averaging 197 nm in diameter, as fat substitutes in creating low-calorie cookies. The cookies' diameter, thickness, and spread ratio were measured, ranging from 46.33 to 57.15 mm, 6.45 to 7.51 mm, and 6.16 to 8.86, respectively. Notably, cookies containing nanoemulsions exhibited a significant increase in the spread ratio compared to the control. The control sample had the highest hardness value at 43.81 N, while the nanoemulsion group had the lowest at 26.98 N. The energy value, which was 508 kcal/100 g in the control group, decreased to 442 kcal/100 g in the group containing the nanoemulsion. The total n-3 fatty acid content in cookies rose from 0.46% in the control cookies to 3.90% in the cookies containing nanoemulsion. Sensory evaluations showed that cookies containing fish ol-loaded nanoemulsion received the highest scores, indicating that the fat reduction did not compromise the desired '' greasy '' sensation. This is especially noteworthy, as it showed that the fat content could be reduced by half without compromising the sensory quality. Utilizing FO-loaded nanoemulsions as a fat replacement in fat-reduced baked goods could provide valuable insights for other food products. The findings have significant implications for the food industry, suggesting that healthier, low-calorie baked goods can be developed without sacrificing physical quality and texture. This approach can cater to the growing market demand for health-conscious food options, potentially leading to new product innovations and enhanced nutritional profiles in a variety of food products.
A Novel Green Tea Extract-Loaded Nanofiber Coating for Kiwi Fruit: Improved Microbial Stability and Nutritional Quality
(Elsevier, 2024) Alav, Aslihan; Kutlu, Nazan; Kina, Erol; Meral, Raciye
This study investigated the use of green tea extract (GTE)-loaded PVA-based nanofibers for preserving fresh-cut kiwi fruit. Scanning Electron Microscopy (SEM) images confirmed that the nanofibers were smooth, uniform, and free from significant defects, with a consistent diameter ranging from 100 to 300 nm. Significant reductions in the total mesophilic aerobic bacteria (TMAB) and yeast and mold counts (TMY) were obtained. On the last day of storage, the TMAB in GTE-coated samples was 1.75 log CFU/g lower than uncoated samples, and the TMY was 2.25 log CFU/g lower (3.50 vs. 5.75 log CFU/g). The kiwi fruit that were coated with nanofibers had higher antioxidant activity, better vitamin C retention (32.3 mg/100 g on day 10 vs. 22.9 mg/100 g for controls), and lower malondialdehyde (MDA) levels (25.3 nmol/kg on day 10 vs. 49.2 nmol/kg for controls), which meant they had less lipid peroxidation and oxidative stress, resulting in improved cellular integrity and extended product shelf life. These findings suggested that GTE-loaded nanofibers could effectively enhance the microbial stability and nutritional quality of fresh-cut kiwi fruit. This study highlighted the potential of using bioactive substanceloaded nanofibers as a novel and effective preservation method for fresh produce, with significant implications for the food industry.
A Novel Solution To Enhance the Oxidative and Physical Properties of Cookies Using Maltodextrin-Based Nano-Sized Oils as a Fat Substitute
(Amer Chemical Soc, 2025) Meral, Raciye; Ekin, Mehmet Mustafa; Ceylan, Zafer; Alav, Aslihan; Kina, Erol
This study investigated the effects of maltodextrin-based nanoemulsions as fat substitutes in cookies, focusing on the oxidative stability and physical properties. Full-fat cookies (control, C) and 50% fat-reduced cookies with nanoemulsions (FC) were produced. The addition of nanoemulsions increased the cookie diameter from 46.3 mm (control) to 56.1 mm and reduced the thickness, resulting in a desirable texture. Initial hardness values (30.3 and 45.8 N) were lower in nanoemulsion samples and remained reduced over a 90 day storage period. Black cumin oil-loaded nanoemulsions provided the lowest peroxide values (1.7, 2.7, and 2.4 mequiv O2/kg), maintaining oxidative stability during storage. Final free fatty acid (FFA) values ranged from 0.23% to 0.44% after storage. Thiobarbituric acid (TBA) values indicated slower lipid oxidation, with values ranging from 1.47 to 2.51 mg MDA/kg on day 0 and increasing to a maximum of 4.13 mg MDA/kg by day 90 in fat-reduced cookies. Among the tested formulations, nanoemulsions enriched with black cumin oil demonstrated the highest effectiveness, yielding enhanced oxidative stability and improved quality characteristics. This study presents an innovative strategy by utilizing maltodextrin-based nanoemulsions containing naturally antioxidant-rich oils as fat replacers, offering a clean-label alternative to improve the oxidative resilience and physical quality of cookies.
Tleablcnn: Brain and Alzheimers Disease Detection Using Attention-Based Explainable Deep Learning and Smote Using Imbalanced Brain Mri
(Ieee-inst Electrical Electronics Engineers inc, 2025) Kina, Erol
Alzheimer's disease (AD) is one of the primary causes of dementia. It degenerates the brain and reduces the activity of individuals by disrupting their memory and physiological functions. A comprehensive examination of specific brain tissue is required in order to accurately diagnose a brain condition using magnetic resonance imaging. The main aim of this research was to develop a rapid and effective technique for detecting healthy persons before the onset of brain tumours, including AD, pituitary tumours, gliomas, and meningiomas. This work presents a lightweight convolutional architecture based on EfficientNet with a Squeeze Attention Block using transfer learning. The proposed approach used lightweight layers with an L2 regularizer, global pooling (2D), and batch normalisation to construct the model, including two dropout layers. This paper utilises the synthetic minority oversampling approach to address the issues of overfitting and imbalanced samples by balancing two significantly imbalanced MRI datasets. We evaluated the efficacy of the minority strategy by doing experiments on both the training set and the entire dataset. The proposed model demonstrates high efficiency, with the risk of overfitting mitigated by applying SMOTE solely to the training data, while the test and validation datasets remain unaffected. The effectiveness of the proposed approach was evaluated using several performance measures and compared with previous published research. We evaluated the proposed approach for explainability via gradient-weighted class activation mapping to comprehend the model's behaviour and its predictions. The proposed framework offers advantages over existing models in terms of computing efficiency, explainability, generalization, and clinical significance.