Classification of Mental Task from EEG Data Using Neural Networks Based on Particle Swarm Optimization

• Main Authors: Chun-Te Wang, 王俊德
• Other Authors: Cheng-Jian Lin
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/nwf52z

碩士 === 朝陽科技大學 === 資訊工程系碩士班 === 94 === The BCI (Brain-Computer Interface) is a system which transforms the brain activity made by different mental task to produce the control signal. The system provides an augmentative communication method to those patients with severe motor disabilities. In this thesis, a method used to classify the electroencephalogram (EEG) of mental task for left hand movement imagination, right hand movement imagination, and word generation is proposed. And we expect the classifying could be used to realize the BCI system. First, the EEG pattern is reduced in a lower dimension and fetched the feature by principle component analysis (PCA). Then, a three layer feed-forward neural network trained by particle swarm optimization (PSO) is used to realize a classifier. The PSO algorithm training the parameters of neural network can avoid some drawbacks of the back-propagation (BP) algorithm like premature converge. The performance demonstration is shown in the result.