Convolution and Wavelet Neural Networks Applied to EEG Brain-Control Interface
Department
Electrical and Computer Engineering
Major
Electrical Engineering
Research Advisor
Wunsch, Donald C.
Advisor's Department
Electrical and Computer Engineering
Funding Source
OURE
Abstract
This project determined the feasibility of applying convolutional and wavelet neural networks to mental imagery classification based EEG Brain Control Interfaces. While statistical techniques like Common Spatial Patterns and Independent Component Analysis have been combined with neural network preprocessing successfully in the past, only recently have EEG researchers begun to take advantage of the flexibility that convolutional neural networks offer for feature extraction and classification. The popularity of these tools for problems such as image recognition has been increasing due to their ability to take advantage of multi-core GPU based computation. Wavelet neural networks offer the potential for reduced training data requirements while maintaining the flexibility for feature extraction and classification across multiple humans.
Biography
Devin Cornell is an undergraduate student in Electrical Engineering. He currently serves as the IEEE Region 5 Student Representative, and has served in several other positions in the IEEE Student Branch at Missouri S&T. He has worked on a microgravity flight research team studying methods of CPR in a space environment for three years and is currently involved with research on methods of using EEG as a Brain-Control Interface in the Missouri S&T Applied Computational Intelligence Laboratory. He has completed two summer internships at Sandia National Labs in Electrical Engineering R&D, and was an exchange student at Universiti Teknologi Petronas in Malaysia. After graduation, he hopes to pursue a PhD in Computational Sociology.
Research Category
Engineering
Presentation Type
Poster Presentation
Document Type
Poster
Location
Upper Atrium/Hall
Presentation Date
15 Apr 2015, 1:00 pm - 3:00 pm
Convolution and Wavelet Neural Networks Applied to EEG Brain-Control Interface
Upper Atrium/Hall
This project determined the feasibility of applying convolutional and wavelet neural networks to mental imagery classification based EEG Brain Control Interfaces. While statistical techniques like Common Spatial Patterns and Independent Component Analysis have been combined with neural network preprocessing successfully in the past, only recently have EEG researchers begun to take advantage of the flexibility that convolutional neural networks offer for feature extraction and classification. The popularity of these tools for problems such as image recognition has been increasing due to their ability to take advantage of multi-core GPU based computation. Wavelet neural networks offer the potential for reduced training data requirements while maintaining the flexibility for feature extraction and classification across multiple humans.
