OPTIMISASI PARAMETER RECURRENT NEURAL NETWORK PADA PENDETEKSIAN ATRIAL FIBRILLATION MENGGUNAKAN ALGORITMA GRID SEARCH

WULANDARI, PUTRI and Nurmaini, Siti (2021) OPTIMISASI PARAMETER RECURRENT NEURAL NETWORK PADA PENDETEKSIAN ATRIAL FIBRILLATION MENGGUNAKAN ALGORITMA GRID SEARCH. Undergraduate thesis, Sriwijaya University.

[thumbnail of RAMA_56201_09011181722073.pdf] Text
RAMA_56201_09011181722073.pdf - Accepted Version
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (2MB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_TURNITIN.pdf] Text
RAMA_56201_09011181722073_TURNITIN.pdf - Accepted Version
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (5MB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_0002085908_01_front_ref.pdf]
Preview
Text
RAMA_56201_09011181722073_0002085908_01_front_ref.pdf - Accepted Version
Available under License Creative Commons Public Domain Dedication.

Download (601kB) | Preview
[thumbnail of RAMA_56201_09011181722073_0002085908_02.pdf] Text
RAMA_56201_09011181722073_0002085908_02.pdf - Accepted Version
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (360kB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_0002085908_03.pdf] Text
RAMA_56201_09011181722073_0002085908_03.pdf - Accepted Version
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (259kB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_0002085908_04.pdf] Text
RAMA_56201_09011181722073_0002085908_04.pdf - Accepted Version
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (911kB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_0002085908_05.pdf] Text
RAMA_56201_09011181722073_0002085908_05.pdf - Accepted Version
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (67kB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_0002085908_06_ref.pdf] Text
RAMA_56201_09011181722073_0002085908_06_ref.pdf - Bibliography
Restricted to Repository staff only
Available under License Creative Commons Public Domain Dedication.

Download (141kB) | Request a copy
[thumbnail of RAMA_56201_09011181722073_0002085908_07_lamp.pdf] Text
RAMA_56201_09011181722073_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

Atrial fibrillation (AF) is an abnormality in the rhythm of the human heartbeat that can cause stroke and heart failure. The process of detecting this disease is done by looking at and analyzing the morphology of the Electrocardiogram (ECG). This research uses the Deep Learning method because it can maximize the use of all information that comes from input, so that the decisions taken are stronger and better. For times series data, the Recurrent Neural Network (RNN) method is suitable for use. The types of RNN used are Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM) and also Gated Recurrent Unit (GRU) because they can overcome the vanishing gradient problem in RNN. There are two cases in this study, namely the use of imbalance data and balance data. The parameter tuning process used Grid Search algorithm by combining parameters such as learning rate, batch size and epoch to found the best model. The best model from each classifier will be tried again by adding a feature extraction process used Convolutional Neural Network (CNN). Based on the experimental results, the model that used the CNN feature extraction process has better results than the model without the feature extraction process. The model that has the best results was CNN-GRU model for data imbalance and data balance. On the imbalance data, the accuracy, sensitivity, specificity, precision, and F1 scores respectively 97.19%, 98.83%, 85.47%, 97.97% and 98.4%. On the balance data, the accuracy, sensitivity, specificity, precision, and F1 scores respectively 93.71%, 91.48%, 96.09%, 96.16%, and 93.76%.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Atrial Fibrillation, Elektrokardiogram, Deteksi, Long Short-Term Memory, Bidirectional Long Short-Term Memory, Gated Recurrent Unit, Convolutional Neural Network, 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: Putri Wulandari
Date Deposited: 09 Jul 2021 04:42
Last Modified: 09 Jul 2021 04:42
URI: http://repository.unsri.ac.id/id/eprint/49527

Actions (login required)

View Item View Item