Optimal EEG feature selection from average distance between events and non-events

Abstract

Biosignal classification systems often have to deal with extraneous features, highly imbalanced datasets, and a low SNR. A robust feature selection/reduction method is a crucial step in this process. Sets of artificial data were generated to test a prototype EEG-based microsleep detection system, consisting of a combination of EEG and 2-s bursts of 15-Hz sinusoids of varied signal-to-noise ratios (SNRs) ranging from 16 to 0.03. The balance between events and non-events was varied between evenly balanced and highly imbalanced (e.g., events occurring only 2% of the time). Features were spectral estimates of various EEG bands (e.g., alpha band power) or ratios between them. A total of 34 features for each of the 16 channels yielded a total of 544 features. Five minutes of EEG from a total of eight subjects were used in the generation of the artificial data. Several feature reduction and classifier structures were investigated. Taking only a single feature corresponding to the maximum of average distance between events and non-events (ADEN) on unbalanced data yielded a phi correlation of 0.94 on the mock data with an SNR of 0.3, compared with a phi coefficient of 0.00 for principal component analysis (PCA). ADEN consistently outperformed alternative system configurations, independent of the classifier utilized. While ADEN's high performance may be due to the nature of the artificial dataset, this simulation has demonstrated strong potential compared to other feature selection/reduction methods.

Evaluation of an English language phoneme-based imagined speech brain computer interface with low-cost electroencephalography

“…Afterward, the percent intensity of each window for 99.95% was calculated. Reflecting other EEG studies, several average power spectral densities bands for the major EEG bands [delta (1–4 Hz), theta (5–8 Hz), alpha (8–12 Hz), beta (13–30 Hz), and gamma (30–100 Hz)] of each feature were calculated using Welch’s method ( LaRocco et al, 2014 , 2020 ). The mean powers of the lower and higher frequency ranges of each EEG band were also calculated (e.g., 8–10 Hz for the lower range of the alpha band).…”

Optimizing Computer–Brain Interface Parameters for Non-invasive Brain-to-Brain Interface

“…Diagnostic protocols, such as those used for therapeutic cases, require the "trial" to be a window long enough to extract meaningful information. Thus, the condition-dependent diagnostic information determines window length (LaRocco et al, 2014). Similarly, the update rate must be sufficient for both real-time response and context-specific features.…”

“…Based on the operational requirements discussed previously, a medical monitoring CBI system is likely to have the highest ITR, owing to its short time window (1.0 s) and rapid update rate (0.1 s). The monitoring system represents a primarily passive high-accuracy system, such as an occupational drowsiness detector (LaRocco et al, 2014). The research CBI is likely to have a high ITR but may be constrained by longer window lengths owing to its large number of classes.…”

The CSP-Based New Features Plus Non-Convex Log Sparse Feature Selection for Motor Imagery EEG Classification

“…The optimal features were further processed for emotion classification using SVM, k-nearest neighbor (k-NN), linear discriminant analysis (LDA), naive Bayes, random forest, deep learning, and four ensembles methods (bagging, boosting, stacking, and voting). The maximum of average distance between events and non-events was used to select optimal EEG features in [ 38 ]. The filter method has certain advantages, such as low computational cost, but it does not consider the correlation between features and is independent of the classifier, so the classification accuracy is not high.…”