Adaptive Brain Interfaces

Brain–Computer Interfaces

“…A few other groups follow machine learning approaches to train the classifier embedded in the BCI. These techniques range from linear classifiers (Babiloni et al, 2001;Obermaier, Müller and Pfurtscheller, 2001), to compact multi-layer perceptrons and Bayesian neural networks (Anderson, 1997;Roberts and Penny, 2000), to variations of LVQ (Kalcher et al, 1996), to local neural classifiers (Millán, 2002;Millán et al, 2002b). Most of these works deal with the recognition of just 2 mental tasks (Roberts and Penny, 2000;Babiloni et al, 2001;Obermaier, Müller and Pfurtscheller, 2001), or report classification errors bigger than 15% for 3 or more tasks (Kalcher et al, 1996;Anderson, 1997).…”

“…Most of these works deal with the recognition of just 2 mental tasks (Roberts and Penny, 2000;Babiloni et al, 2001;Obermaier, Müller and Pfurtscheller, 2001), or report classification errors bigger than 15% for 3 or more tasks (Kalcher et al, 1996;Anderson, 1997). An exception is Millán's approach that achieves error rates below 5% for 3 mental tasks, but correct recognition is 70% (Millán, 2002;Millán et al, 2002b). Obermaier, Müller and Pfurtscheller (2001) reports on a single disabled person who, after several months of training, has reached a performance level close to 100%.…”

“…These classification rates, together with the number of recognizable tasks and duration of the trials, yield bit rates from approximately 0.15 to 2.0. Some approaches are based on a mutual learning process where the user and the brain interface are coupled together and adapt to each other (Roberts and Penny, 2000;Obermaier, Müller and Pfurtscheller, 2001;Millán, 2002;Millán et al, 2002b). This should accelerate the training time.…”

Brain-Computer Interface Based on Generation of Visual Images

“…Basic research has revealed correlations between brain signals and mental states [1], [2], [3], [4], [5]). This provides a variety of brain signals which might be used for BCI design [1].…”

“…One approach for BCI design is based on the discrimination of EEG patterns related to different mental states [4], [11], [12]. In this approach the subject is requested to perform different mental tasks.…”