International Journal of Advances in Computer Science and Its Applications
Author(s) : JUHI AGARWAL, R. H. GOUDAR
Brain Computing Interface (BCI) provides the direct link between brain and computer device. People do not need any muscle movement. Brain states can be detected and translated into actions. Two basic requirements of brain and computing interface are the features useful to distinguish several kinds of brain states and methods for classification of signals.BCI have three types: invasive, partially invasive and noninvasive BCI, but non invasive is a safe technique.EEG is the most popular non invasive technique. After acquisition of signals, the feature extraction and classification methods are performed. These methods will play the main role in BCI system’s output. If the misclassification is performed, then the error or wrong command will generate. Currently we have so many methods available for the classification like Linear discriminate analysis (LDA), Support vector machine (SVM), multiple layer perception (MLP), bayes quaderatic etc. but there are many challenges and issues in BCIs. The challenges are Adaptation and learning is very tuff, BCI systems cannot be used autonomously by disabled people, Hard to adapt changes occurring continuously EEG and issues are like BCI requires extensive training from several hours to several months, BCI invasive technique can lead to damage of tissues, BCI provides low transfer-rate(5-25 bits/min).In the real life there are so many applications of BCI. So BCI is very useful technique for future, but BCI need some more good technology for capturing the EEG signals and handle the noise and methods for classification for improving the output of BCI. This paper discusses the architecture, tools and techniques, issues and challenges, research directions in Brain Computer Interface for setting Brain into action.