International Journal on Advanced Science, Engineering and Information Technology, Vol. 13 (2023) No. 4, pages: 1486-1491, DOI:10.18517/ijaseit.13.4.18925

An Analysis of Several Optimizers on CNNSVM and CNNRF for COVID–19 Chest X–ray Images

Zuherman Rustam, Jane Eva Aurelia, Fevi Novkaniza, Sri Hartini, Rahmat Hidayat, Mostafa Ezziyyani

Abstract

COVID–19 is a new type of ailment caused by the strenuous acute respiratory syndrome, namely SARS–CoV-2, also frequently well–known as the Coronavirus. An early tendency of COVID–19 for some sufferers can cause no symptoms at all as no experience is referred to as asymptomatic confirmation cases, yet these sufferers can still transmit COVID–19 to other people. Therefore, the authors developed a program using Machine Learning that sustains data to be analyzed based on the input served under the proposed methods of Convolutional Neural Network–Support Vector Machine (CNNSVM) and Convolutional Neural Network–Random Forest(CNNRF), along with several optimizers to be compared. Convolutional Neural Networks is a deep learning algorithm that can train large data sets with millions of parameters and has attracted attention in various fields that are commonly used for the classification and detection of Convolution in Neural Networks. In amalgamation with Support Vector Machines, a technique that uses two vectors to form a dividing line or side and fairly high accuracy,y random forests classification. In the manner of image data obtained from ChestX–ray images of people with COVID–19 from the Italian Society of Medical and Interventional Radiology (SIRM), a total of 1750 observations consisting of 1000 data for COVID¬–19 images and 750 data for non-COVID–19 images. This research aims to determine which optimizer is better for analyzing COVID–19 ChestX–ray images by evaluating both methods. Hopefully, both methods can provide higher accuracy for future studies with wider databases to provide better results for analyzing different ailments.

Keywords:

COVID–19; ChestX–ray; machine learning; convolutional neural networks; support vector machines; random forests

Viewed: 251 times (since abstract online)

cite this paper     download