Mahalanobis distance-based classifiers are able to recognize EEG patterns by using few EEG electrodes

Abstract

Abstract-In this paper, we explore the use of quadratic classifiers based on Mahalanobis distance to detect EEG patterns from a reduced set of recording electrodes. Such classifiers used the diagonal and full covariance matrix of EEG spectral features extracted from EEG data. Such data were recorded from a group of 8 healthy subjects with 4 electrodes, placed in C3, P3, C4, P4 position of the international 10-20 system. Mahalanobis distance classifiers based on the use of full covariance matrix are able to detect EEG activity related to imagination of movement with affordable accuracy (average score 98%). Reported average recognition data were obtained by using the cross-validation of the EEG recordings for each subject. Such results open the avenue for the use of Mahalanobis-based classifiers in a brain computer interface context, in which the use of a reduced set of recording electrodes is an important issue.

Comparison of Classifiers and Statistical Analysis for EEG Signals Used in Brain Computer Interface Motor Task Paradigm

“…In mathematical terms, the Mahalanobis distance is equal to the Euclidean distance when the covariance matrix is the unit matrix. The use of the Mahalanobis distance removes several of the limitation of linear classifiers based on Euclidean metric, since it automatically account for the scaling of the coordinate axes, as well as for the correlation between the different considered features [9]. Mahalanobis classifier is simple but at the same time robust and leads to good results, as shown in [10].…”

“…The oscillatory activity in the EEG is classified according to frequency bands or rhythms: Delta (1-4 Hz), Theta (4-8 Hz), Alpha and Mu (8)(9)(10)(11)(12), Beta (13-25 Hz), Gamma (25-40 Hz) [2]. Mu rhythm (8)(9)(10)(11)(12) Hz) is affected by movements or movement imagery.…”

A High Performance MEG Based BCI Using Single Trial Detection of Human Movement Intention

“…The Mahalanobis Distance Classifier had proved effective for classification in previous studies [27,57]. It was further optimized using GA-based feature extraction method.…”

Using a Mahalanobis-Like Distance to Train Radial Basis Neural Networks

“…This distance can be used to improve prediction accuracy in those learning systems that use distances [9]. However, the Mahalanobis distance is independent of the learning system used and of the error produced on the training data, because it is computed from the points in the data set only (more specifically, it is computed from the variance-covariance matrix).…”