Saddam, Muhammad (2017) Klasifikasi Eeg Alkoholik Menggunakan Wavelet Packet Decomposition, Principal Component Analysis, Dan Kombinasi Genetic Algorithm Dan Neural Network. Undergraduate thesis, InstitutTeknologi SepuluhNopember.
Preview |
Text
5113100187-Undergraduate_Theses.pdf - Published Version Download (2MB) | Preview |
Abstract
Alkoholisme adalah gangguan yang ditandai oleh konsumsi berlebihan dan ketergantungan pada alkohol. Terdapat bermacam cara untuk mendeteksi apakah seorang pasien telah kecanduan alkohol, salah satunya dengan deteksi otak menggunakan Electroencephalographic (EEG).
Sinyal EEG secara luas dan klinis digunakan untuk melakukan deteksi gangguan otak pada dunia kesehatan. Akan tetapi, sinyal yang dihasilkan oleh EEG perlu dipersiapkan untuk dilakukan proses agar dapat mendeteksi kelainan otak secara otomatis. Oleh karena itu, perlu adanya metode praproses untuk ekstraksi fitur yang tepat agar mendapatkan karakteristik yang tersimpan secara implisit dari sinyal EEG tersebut.
Tugas Akhir ini akan mengimplementasikan metode Wavelet Packet Decomposition (WPD) untuk ekstraksi fitur, Principal Component Analysis (PCA) untuk reduksi dimensi, dan Neural Network yang dioptimasi dengan metode Genetic Algorithm dalam pencarian bobot dan bias optimal untuk klasifikasi data alkoholik dan normal. Data uji coba yang digunakan dibagi menjadi 2 yaitu, dataset 1 terdiri dari 60 training dan 40 testing serta dataset 2 terdiri dari 120 training dan 40 testing. Berdasarkan hasil uji coba, rata-rata akurasi terbaik didapatkan dari dataset 1 sebesar 94.00% dengan dekomposisi 3 level, penggunaan fitur 30%, dan klasifikasi menggunakan Kombinasi Neural Network dan Genetic Algorithm dengan learning rate 0.1 dan nilai alpha pada proses crossover 0.9.
===============================================================================================
Alcoholism is a disorder characterized by excessive consumption and dependence on alcohol. There are various ways to detect whether a patient is addicted to alcohol, one of them by brain detection using electroencephalographic (EEG).
EEG signals are widely and clinically used for detection of brain disorders in the medical world. However, the signals generated by the EEG should be prepared to do further processing to detect brain abnormalities automatically. Initially, preprocessing method is needed for proper feature extraction, in order to obtain characteristics that are stored implicitly in the EEG signal.
This undergraduate thesis implements Wavelet Packet Decomposition (WPD) method for feature extraction, Principal Component Analysis (PCA) for dimension reduction, and Back Propagation Neural Network optimized with Genetic Algorithm to get optimal weights and biases for alcohol addiction classification. The EEG data used are divided into two: the first dataset consists of 60 training data and 40 testing data, and the second dataset consists of 120 training data and 40 testing data. Based on the experiment results, the first dataset gives best performance with 94.00% accuracy with decomposition of 3 levels, taking 30% of the features, and classification using Neural Network and Genetic Algorithm with learning rate of 0.1 and crossover alpha value of 0.9.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSIf 621.398 1 Sad k |
| Uncontrolled Keywords: | Alkoholism, EEG, Wavelet Packet Decomposition, Principal Component Analysis, Neural Network, Genetic Algorithm |
| Subjects: | Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) R Medicine > R Medicine (General) T Technology > T Technology (General) |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | - MUHAMMAD SADDAM |
| Date Deposited: | 24 Oct 2017 03:14 |
| Last Modified: | 05 Mar 2019 01:58 |
| URI: | http://repository.its.ac.id/id/eprint/41697 |
Actions (login required)
 |
View Item |