Open AccessRECOGNITION OF COGNITIVE POTENTIALS TO THE TARGET STIMULI IN THE BRAIN-COMPUTER INTERFACE ON THE BASIS OF THE ENSEMBLE OF CLASSIFIERS
Author(s) -
D.A. Kirjanov,
Arline Kaplan
Publication year - 2016
Publication title -
nauka i innovacii v medicine
Language(s) - English
Resource type - Journals
eISSN - 2618-754X
pISSN - 2500-1388
DOI - 10.35693/2500-1388-2016-0-3-28-32
Subject(s) - brain–computer interface , computer science , linear discriminant analysis , electroencephalography , pattern recognition (psychology) , artificial intelligence , interface (matter) , support vector machine , task (project management) , speech recognition , machine learning , psychology , neuroscience , management , bubble , maximum bubble pressure method , parallel computing , economics
Background. A number of studies have been done on detection of human visual attention focus by means of P300 brain-computer interfaces (BCI). However, the performance of interfaces on P300 is still low, since this technique requires the repeated presentation of target and non-target stimuli. There are some indications that it is appropriate to use ensembles of classifiers to improve the accuracy of recognition of multidimensional objects. The goal of the present study was to verify the feasibility of application of ensembles of classifiers to speed up the work of the BCI P300. Methods. The study involved 22 subjects, whose task was to closely monitor the highlights of target objects on the computer screen, presented as 8 triangles located in a circle (angle of 7.7 degrees). Single classifiers and ensembles of classifiers based on linear discriminant of Fisher were used to detect the target responses in the EEG. Results. The use of the ensemble of classifiers provided almost the same accuracy of algorithmic choice of target reactions, EEG in 78-80%, as compared with the use of single classifiers, but with two times smaller number of repetitions of the test stimuli and, therefore, faster detection of the target reactions of the EEG. Conclusions. This work implies that the P300 BCI with the participation of the ensemble of classifiers can be used to build high-speed communication systems for both the stroke patients and healthy people in special circumstances for additional alarm at the inability to use speech.