Klasifikasi Aktivitas Mental Dari Data Eeg Menggunakan Neural Network Dan Fuzzy Particle Swarm Optimization Dengan Cross-Mutated Operator - Classification Of Mental Activities From Eeg Data Using Neural Network And Fuzzy Particle Swarm Optimization With Cross-Mutated Operator

Sakur, Stendy B (2015) Klasifikasi Aktivitas Mental Dari Data Eeg Menggunakan Neural Network Dan Fuzzy Particle Swarm Optimization Dengan Cross-Mutated Operator - Classification Of Mental Activities From Eeg Data Using Neural Network And Fuzzy Particle Swarm Optimization With Cross-Mutated Operator. Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 5113201044-Master Thesis.pdf]
Preview
Text
5113201044-Master Thesis.pdf - Published Version

Download (7MB) | Preview
[thumbnail of 5113201044-Paper.pdf]
Preview
Text
5113201044-Paper.pdf - Published Version

Download (1MB) | Preview
[thumbnail of 5113201044-Presentation.pdf]
Preview
Text
5113201044-Presentation.pdf - Published Version

Download (1MB) | Preview

Abstract

Brain-computer interface (BCI) is a system that can transform electrical
activities in the brain to different mental activities into the control signals.
Electroencephalogram (EEG) based BCI is the result of the measurement of the
electrical activity of the brain waves generated by the cerebral cortex. In general, a
neural network is widely used as a method of multi-class classification, but very
slow in reaching convergence or the solution can be trapped in a local minimum.
Evolutionary algorithm is proposed to optimize the search weights of the neural
network, including the Particle Swarm Optimization (PSO). The fundamental
issue is the speed of PSO is less stable that need improvement in all components
of the particle. The component consists of the weight of the inertia of the previous
speed, knowledge of individual and group knowledge. The component has a great
influence in achieving the level of convergence, so it is important to note at the
time of optimization weighting.
The purpose of this study is to propose a new strategy on methods
Improved Particle Swarm Optimization (IPSO) which uses Modified Evolutionary
Direction Operator with Adaptive Inertia weights. Where Fuzzy Inference System
is used as adaptive inertia weight in optimizing the weighting of Neural Network.
The strategy is expected to improve the accuracy.
The results show, for classification first subject is 54.20%, second subject
of 58.40% and 50.80% for subject third with the average accuracy of 54.48%.
Where the increase for first subject is 1.26%, second subject is 12.63% and the
third subject 0.80%. Thus the proposed method is Fuzzy Modified Evolutionary
Direction operator with cross-mutated operation is better than the previous
method.
========================================================================================================================
Brain-computer interface (BCI) is a system that can transform electrical
activities in the brain to different mental activities into the control signals.
Electroencephalogram (EEG) based BCI is the result of the measurement of the
electrical activity of the brain waves generated by the cerebral cortex. In general, a
neural network is widely used as a method of multi-class classification, but very
slow in reaching convergence or the solution can be trapped in a local minimum.
Evolutionary algorithm is proposed to optimize the search weights of the neural
network, including the Particle Swarm Optimization (PSO). The fundamental
issue is the speed of PSO is less stable that need improvement in all components
of the particle. The component consists of the weight of the inertia of the previous
speed, knowledge of individual and group knowledge. The component has a great
influence in achieving the level of convergence, so it is important to note at the
time of optimization weighting.
The purpose of this study is to propose a new strategy on methods
Improved Particle Swarm Optimization (IPSO) which uses Modified Evolutionary
Direction Operator with Adaptive Inertia weights. Where Fuzzy Inference System
is used as adaptive inertia weight in optimizing the weighting of Neural Network.
The strategy is expected to improve the accuracy.
The results show, for classification first subject is 54.20%, second subject
of 58.40% and 50.80% for subject third with the average accuracy of 54.48%.
Where the increase for first subject is 1.26%, second subject is 12.63% and the
third subject 0.80%. Thus the proposed method is Fuzzy Modified Evolutionary
Direction operator with cross-mutated operation is better than the previous
method.

Item Type: Thesis (Masters)
Additional Information: RTIf 621.398 1 Sak k
Uncontrolled Keywords: Brain-computer interface, Electroencephalogram, Artificial Neural Network, Particle Swarm Optimization, Fuzzy Inertia Weight, Evolutionary Direction Operator, Cross-mutated Operation, Brain-computer interfaces, Electroencephalogram, arificial Neural Network, Particle Swarm Optimization, Fuzzy Inertia Weight, Evolutionary Direction Operator, Cross-mutated operation
Subjects: Q Science > QP Physiology > Q376.5 Electroencephalography (EEG)
Divisions: Faculty of Information Technology > Informatics Engineering
Depositing User: ansi aflacha
Date Deposited: 21 Nov 2019 09:00
Last Modified: 21 Nov 2019 09:00
URI: http://repository.its.ac.id/id/eprint/71958

Actions (login required)

View Item View Item