Musa, Saiful Bahri (2017) Klasifikasi Emosi Sinyal EEG berdasarkan Empirical Mode Decomposition dan Wavelet Packet Decomposition menggunakan Logarithmic Learning For Generalized Classifier Neural Network. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5114201051-Master-Theses.pdf - Published Version Download (2MB) | Preview |
Abstract
Emosi manusia berdasarkan aktivitas otak dapat menghasilkan gelombang elektrik yang sangat kecil. Dengan electroencephalography (EEG) dapat diperoleh data rekaman aktivitas otak dari sejumlah channel – channel berdasarkan pemantauan electrophysiological. Masing-masing channel memberikan respons yang berbeda-beda saat mendapatkan stimulus emosi. untuk mengenali emosi manusia berdasarkan aktivitas gelombang otak, dibutuhkan penguraian atau perubahan sinyal yang dapat diartikan sebagai nilai penting dalam menentukan emosional manusia.
Penelitian ini mengusulkan sebuah kerangka klasifikasi emosi manusia dari data sinyal EEG menggunakan metode Logarithmic learning for Generalized Classifier Neural Network (L-GCNN), sinyal yang diklasifikasi adalah hasil analisis dari metode EMD untuk proses shifting dengan cara menguraikan rangkaian waktu sinyal menjadi sejumlah Intrinsic Mode Functions (IMF) dan metode WDP dengan membentuk sinyal menjadi subband - subband yang terdiri dari approximation dan detail. Selanjutnya masing-masing subband akan di hitung berdasarkan perhitungan statistik logaritma sehingga membentuk data fitur.
Dari hasil uji coba berdasarkan pemilihan channel-channel pada area dahi, area telinga kiri dan kanan, serta area tengkuk (leher bagian belakang) diperoleh rata-rata akurasi adalah 86.94 % untuk skenario B dan skenario A mendapatkan rata-rata akurasi 68.45 %. Dari hasil uji coba diperoleh kesimpulan bahwa skenario B lebih baik dari skenario A dalam mengklasifikasi emosi manusia berdasarkan analisis dekomposisi dengan menerapkan hierarki ke 6 sampai hierarki ke 8 pada metode WPD.
=================================================================
Human emotion based on brain activity can produce a very small electrical waves. With electroencephalography (EEG) the recording data of brain activity from several channels can be obtained by electrophysiological monitoring. Each channel provides different response when receive the emotion stimulus. To identify the human emotion based on brain wave activity, decomposition or signal change which can be interpreted as important value in determining human emotion is needed.
This study proposes a classification framework of human emotion from EEG signals data using logarithmic learning for Generalized Classifier Neural Network (L-GCNN), the classified signal is a analysis result from Empirical Mode Decomposition (EMD) method for shifting process by decomposing the signal time-series into Intrinsic Mode Function (IMF) and Wavelet Packet Decomposition (WDP) method by forming the signal into sub-bands which is consist approximation and detail. Furthermore, each sub-bands will be calculated based on logarithm statistical calculations to form a feature data.
From the evaluation results based on channels selection on the forehead area, the left and right ears area and also the nape area (rear parts of the neck), the obtained average accuracy was 94% for B scenario and 84% for A scenario. From the evaluation results it is concluded that both A and B scenario managed to classify the human emotion based on decomposition analysis by applying the 6th to 8th hierarchy on the WPD method.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Electroencephalogram; DEAP; Empirical Mode Decomposition; Wavelet Packet Decomposition; Logarithmic learning for General Classifier Neural Network (L-GCNN |
| Subjects: | B Philosophy. Psychology. Religion > B Philosophy (General) Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4050 Electronic information resources |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55101-(S2) Master Thesis |
| Depositing User: | SAIFUL BAHRI MUSA |
| Date Deposited: | 02 Mar 2017 02:37 |
| Last Modified: | 05 Mar 2019 07:55 |
| URI: | http://repository.its.ac.id/id/eprint/2459 |
Actions (login required)
 |
View Item |