IMPLEMENTASI ALGORITMA CNN DENGAN ARSITEKTUR RESNET-50 UNTUK KLASIFIKASI MALARIA

Malaria is an infectious disease transmitted through the bite of a female anopheles mosquito that contains plasmodium in its body. This research aims to develop a classification system to detect red blood cells infected with malaria. This system uses the Convolutional Neural Network (CNN) method with ResNet-50 architecture. The data collected is in the form of malaria disease image data totaling 27,558. All images entered into the ResNet-50 layer must be resized. After training and testing the model, the next step is the calculation of the Confusion Matrix. The training process is divided into 6 stages, namely the SGD optimizer with a learning rate of 0.01, 0.001, 0.0001 and the Adam optimizer with a learning rate of 0.01, 0.001, and 0.0001. Based on the comparison, a significant performance comparison is obtained between the SGD and Adam optimizer models at various learning rate values. The SGD optimizer provides better results with more optimal loss and accuracy values than the Adam optimizer, especially at a learning rate of 0.01. The deep learning method with ResNet-50 architecture is able to detect malaria from red blood cell images with a good level of accuracy and can be an alternative solution in diagnosing malaria.