SMOTE Variants in Pneumonia Detection: Comparative Model Analysis Pnömoni Tespitinde SMOTE Varyantlari: Karsilastirmali Model Analizi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11636463
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Artificial Neural Networks, Data Augmentation, Pneumonia, SMOTE, SMOTE-TOMEK, X-Ray
- Ankara Yıldırım Beyazıt Üniversitesi Adresli: Evet
Özet
This study analyzes the effects of SMOTE-based data augmentation techniques on different machine learning algorithms for a dataset of imbalanced X-ray images. For a dataset consisting of two classes (healthy and pneumonia), five different feature families commonly used in the literature were calculated after a preprocessing step involving basic standardization. Three different SMOTE methods were used to address the imbalance problem between the classes, and their effects on different machine learning algorithms were examined. Based on the models performance, the best results were achieved using the SMOTE-TOMEK and Borderline-SMOTE algorithms. When the generated data was analyzed statistically, the Borderline-SMOTE algorithm obtained the most accurate results.