OPTIMISASI PARAMETER CONVOLUTIONAL NEURAL NETWORK 1-DIMENSI PADA KLASIFIKASI PENYAKIT JANTUNG MENGGUNAKAN ALGORITMA GRID SEARCH

KHOTIMAH, ALNA YOPA and Nurmaini, Siti (2021) OPTIMISASI PARAMETER CONVOLUTIONAL NEURAL NETWORK 1-DIMENSI PADA KLASIFIKASI PENYAKIT JANTUNG MENGGUNAKAN ALGORITMA GRID SEARCH. Undergraduate thesis, Sriwijaya University.

	Text RAMA_56201_09011181722003.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (6MB) | Request a copy
	Text RAMA_56201_09011181722003_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (15MB) | Request a copy
Preview	Text RAMA_56201_09011181722003_0002085908_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (1MB) | Preview
	Text RAMA_56201_09011181722003_0002085908_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (686kB) | Request a copy
	Text RAMA_56201_09011181722003_0002085908_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (730kB) | Request a copy
	Text RAMA_56201_09011181722003_0002085908_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (4MB) | Request a copy
	Text RAMA_56201_09011181722003_0002085908_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (194kB) | Request a copy
	Text RAMA_56201_09011181722003_0002085908_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (320kB) | Request a copy
	Text RAMA_56201_09011181722003_0002085908_07_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy

Abstract

Heart disease is a condition where the heart does not work normally so that it can affect the structure and function of the heart itself. One of the medical tests that can be done to detect heart disease is an electrocardiogram (EKG). Errors during diagnosis often occur when the ECG is analyzed manually. However, in recent years the computational processing has been done a lot. In the research conducted, Convolutional Neural Network (CNN) 1-dimensional architecture is able to learn features directly so as to prevent the loss of important features and can improve accuracy. In addition, the optimization method uses a grid search algorithm to optimize the parameters to improve the performance of the proposed architecture. The ECG signal databases used are The PTB Diagnosis and BIDMC Congestive Heart Failure. In this study, classification of heart disease was carried out in 4 classes and 6 classes were tested to obtain the best combination model of the parameters of batch size, learning rate, and epoch. The best combination of parameters in 6 clases is a batch size of 16, a learning rate of 0.0001 and an epoch of 100 with a performance of 99.37% accuracy, 91.91% sensitivity, 99.18% specificity, 95.58% precision, and an F1-score of 93.67%. Then the model was tested using K-Fold with the best model on the 9th fold with an accuracy of 99.50%, sensitivity 92.28%, specificity 99.38%, precision 96.57%, and F1-score of 94.30%.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Elektrokardiogram, Klasifikasi, Optimisasi, Convolutional Neural Network, dan Grid Search.
Subjects:	Q Science > Q Science (General) > Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation.
Divisions:	09-Faculty of Computer Science > 56201-Computer Systems (S1)
Depositing User:	Alna Yopa Khotimah
Date Deposited:	09 Jul 2021 04:38
Last Modified:	09 Jul 2021 04:38
URI:	http://repository.unsri.ac.id/id/eprint/49525

Actions (login required)

View Item