Study of brain computer Interface using soft computing techniques

Abstract

Brain Computer Interface (BCI) is being widely studied as a communication newlinesolution for physically impaired people to continue day-to-day operations without newlineusing muscular activity. The key challenge in BCI is the classification of the brain newlineactivity patterns according to the activity the users wish to perform and translate newlinethe same into commands, which can be used by a computer or electronic device. newlineMany of the published results are related to the investigation and evaluation of newlineclassification algorithms because of the increased interest for Electroencephalogram newline(EEG)-based BCI. newlineIn EEG (Electro-encephalogram) signals, there would be a cluster of newlinefeatures, and it is vital to extract the useful features from them. Identifying and newlineextracting good features from the signals is a crucial step in the design of BCI. The newlinestudies suggest two important areas for successful implementation of BCI in real newlinelife. The first is being the removal of artifacts and extracting features in the newlinefrequency domain. The second is the classification algorithm to improve the newlineclassification accuracy. Feature extraction in the frequency domain using newlineWavelets. Wavelets are considered as EEG signals are non-stationary. The primary newlineobjectives of the research are: newlineand#61623; To aptly classify ECoG images by using the techniques of classification newlineand#61623; To choose the features that are best for the ECoG image using the selection techniques newlinelike Mutual Information and Information Gain. newlineand#61623; To select the best image by employing the Genetic Algorithm feature selection newlineand#61623; To choose features with the proposed GA and its classifier. newlineand#61623; To be able to propose techniques of feature extraction along with classifiers for newlineclassifying images efficiently. newlineand#61623; To propose the GA-based feature selection with Support Vector Machine (SVM), newlineRandom Forest (RF), and Logistic Regression (LR) classifiers. newline newline