Adaptive Classification of Mental States for Asynchronous Brain Computer Interfaces

Brain Computers Interfaces (BCI) are emerging as a new means of communication, aiming to make a direct link between the brain and an external device, bypassing conventional motor outputs, such as peripheral nerves and muscles. A BCI extracts features from a brain signal and classifies them in order to interpret them in terms of the user's volition. For communication to be effective, the computer has to provide feedback to the user allowing him/her to judge how the brain activity is being classified and interpreted. Similarly, the user must produce patterns of brain activity which can easily be learned and recognized by the computer. Here, we describe a method for selecting mental tasks that are best classified by a subject using support vector machines (SVM).