Perbandingan penerapan optimasi SGDM dan Adam pada model CNN dengan arsitektur VGG19 dan ResNet-50 dalam memprediksi penyakit paru-paru pneumonia

Abstract

Pneumonia is a leading cause of death among children under five, accounting for 14% of fatalities. Chest X-ray analysis is a key method for diagnosis, but many developing countries have only one radiologist per million people, making timely detection difficult. To address this challenge, Convolutional Neural Networks (CNN) offer a viable solution due to their ability to analyze visual data efficiently. This study evaluates two CNN architectures, VGG19 and ResNet-50, considering their effectiveness in pneumonia detection. Both models were trained using two different optimizers, SGDM and Adam, to determine the best combination for accurate classification. Results using test data indicate that VGG19 with the Adam optimizer achieves the highest accuracy at 90%, surpassing other models which recorded 62%, 77%, and 84% without overfitting. This highlights the potential of artificial intelligence driven diagnostic tools in bridging healthcare gaps and improving pneumonia detection in resource-limited settings.

Keywords: Classification, CNN, Optimizer, Pneumonia
MSC2020: 62