Performance Improvement of Near-Infrared Spectroscopy-Based Brain-Computer Interface Using Regularized Linear Discriminant Analysis Ensemble Classifier Based on Bootstrap Aggregating

Abstract

Ensemble classifiers have been proven to result in better classification accuracy than that of a single strong learner in many machine learning studies. Although many studies on electroencephalography-brain-computer interface (BCI) used ensemble classifiers to enhance the BCI performance, ensemble classifiers have hardly been employed for near-infrared spectroscopy (NIRS)-BCIs. In addition, since there has not been any systematic and comparative study, the efficacy of ensemble classifiers for NIRS-BCIs remains unknown. In this study, four NIRS-BCI datasets were employed to evaluate the efficacy of linear discriminant analysis ensemble classifiers based on the bootstrap aggregating. From the analysis results, significant (or marginally significant) increases in the bitrate as well as the classification accuracy were found for all four NIRS-BCI datasets employed in this study. Moreover, significant bitrate improvements were found in two of the four datasets.

Subject-Independent Functional Near-Infrared Spectroscopy-Based Brain–Computer Interfaces Based on Convolutional Neural Networks

“…Recently, Shin and Im (2020) demonstrated that ensemble of weak classifiers resulted in a better classification accuracy than that of a single strong classifier. Based on this work, the ensemble of regularized LDA based on bootstrap aggregating (Bagging) algorithm was employed to validate the performance of subjectindependent fNIRS-based BCI.…”

“…In this study, the ensemble classifier was implemented using the MATLAB "fitcensemble" function. According to the previous study (Shin and Im, 2020), the number of weak classifiers, fraction of training set to resample, and gamma value for regularized LDA were set to 50, 100%, and 0.1, respectively.…”

Global research on artificial intelligence-enhanced human electroencephalogram analysis

“…First, from the algorithm perspective, most studies were concerned with classification issues in the field of EEG analysis (e.g., [230][231][232][233][234]), for example, human mental emotion classification [230], classification of forearm movement imagery [231], detection of acute pain signals [232], sleep stage classification [235], classification of individuals into a normal group and one with particular diseases [236], and classification of repeating stimuli as either old or new [237]. Furthermore, ensemble classifiers are more effective than a single strong learner [238]. Thus, it is suggested that scholars should Second, machine learning and deep learning have been widely applied to EEG data analysis, particularly CNNs (e.g., [239][240][241][242][243][244][245]).…”

Regional analysis of cerebral hemodynamic changes during the head-up tilt test in Parkinson’s disease patients with orthostatic intolerance

“…Ensemble methods for NIRS data improve machine learning performance through the combination of several weak learners. 40 Decision tree classification randomly samples the input data (i.e., bootstrapping) and tests the out-of-bag samples. This randomization process decorrelates individual decision trees to prevent overfitting.…”