Klasifikasi Electroencephalogram (EEG) Motor Imagery Dengan Fitur Differential Asymmetry Pada Support Vector Machine (SVM)

Putranto, Yulianto Tejo (2023) Klasifikasi Electroencephalogram (EEG) Motor Imagery Dengan Fitur Differential Asymmetry Pada Support Vector Machine (SVM). Doctoral thesis, Institut Teknologi Sepuluh Nopember.

Text 07111560010019-Dissertation.pdf - Accepted Version Restricted to Repository staff only until 1 April 2025. Download (3MB) | Request a copy

Abstract

Selama bertahun-tahun penelitian bidang Brain-Computer Interface (BCI) telah difokuskan pada keinginan untuk menyediakan para penyandang disabilitas kemudahan dan kemampuan untuk berinteraksi dengan lingkungannya. Dalam beberapa tahun terakhir, penelitian BCI telah merambah ke aplikasi yang diperuntukkan bagi orang normal untuk meningkatkan kualitas hidup atau mendapatkan keuntungan komersial bagi suatu komunitas atau target group. Pada penelitian ini dirancang sistem BCI berbasis EEG, khususnya untuk mengenali aktivitas motor imagery untuk diterapkan dalam perangkat keras atau perangkat lunak yang bersifat interaktif.
Sistem BCI menuntut akurasi dan kecepatan respon yang tinggi. Permasalahan ini dicoba dipecahkan dengan mengembangkan ekstraksi fitur yaitu differential asymmetry berdasarkan nilai selisih antara hasil pengukuran sinyal otak belahan kiri dan kanan. Fitur-fitur statistik dari hasil dekomposisi sinyal EEG motor imagery menggunakan transformasi wavelet diskrit dan dekomposisi mode empiris dimodifikasi dengan mengambil nilai selisihnya untuk dijadikan fitur baru. Sebagai pengklasifikasi digunakan Support Vector Machine (SVM).
Hasil penelitian menunjukkan peningkatan nilai akurasi dari pengklasifikasi dengan menerapkan fitur differential asymmetry dibandingkan tanpa menerapkan fitur differential asymmetri. Dari tiga dataset yang diteliti, dataset I mengalami kenaikan nilai akurasi rata-rata sebesar 30,33%, dataset II sebesar 6,54% dan dataset III sebesar 23,17%. Hasil akurasi dengan menggunakan SVM untuk dataset I, dataset II dan dataset III berturut-turut: 91,70%, 66,91% dan 93,16%.
=================================================================================================================================
Through the years, research in Brain-Computer Interface (BCI) has focused on providing convenience and ability for disabilities people to interact with the environment. In recent years, BCI research has been enlarge into applications intended for normal people to improve life quality or obtain commercial advantage for a community or group target. In this study, an EEG-based BCI system was designed, especially for recognizing motor imagery activity to be applied in games.
The BCI system demands accuracy and response speed. This problem will be solved by developing a feature extraction, namely differential asymmetry based on the value of the difference between the measurement results of the left and right brain hemisphere signals. Statistical features from the decomposition of motor imagery EEG signals using discrete wavelet transform (DWT) and empirical mode decomposition (EMD) are modified by taking those difference values was as a new feature. As a classifier, Support Vector Machine (SVM) is used.
The results showed an increase in the accuracy value of the classifier by applying the differential asymmetry feature compared to without applying the differential asymmetry feature. Of the three datasets studied, dataset I experienced an average increase in accuracy by 30.33%, dataset II by 6.54% and dataset III by 23.17%. Accuracy results using SVM for dataset I, dataset II and dataset III are respectively: 91.70%, 66.91% and 93.16%.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	Kata Kunci: Brain-Computer Interface (BCI), EEG motor imagery, dekomposisi mode empiris, differential asymmetry, SVM, transformasi wavelet diskrit ========================================================== Keywords: Brain-Computer Interface (BCI), differential asymmetry, discrete wavelet transform (DWT), EEG motor imagery, empirical mode decomposition (EMD), SVM
Subjects:	T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5102.9 Signal processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20001-(S3) PhD Thesis
Depositing User:	Yulianto Tejo Putranto
Date Deposited:	13 Feb 2023 06:13
Last Modified:	13 Feb 2023 06:13
URI:	http://repository.its.ac.id/id/eprint/97190

Actions (login required)

View Item