ERP-based Brain-computer Interfaces Research Articles

A comprehensive dataset for home appliance control using ERP-based BCIs with the application of inter-subject transfer learning.

Brain-computer interfaces (BCIs) have a potential to revolutionize human-computer interaction by enabling direct links between the brain and computer systems. Recent studies are increasingly focusing on practical applications of BCIs-e.g., home appliance control just by thoughts. One of the non-invasive BCIs using electroencephalography (EEG) capitalizes on event-related potentials (ERPs) in response to target stimuli and have shown promise in controlling home appliance. In this paper, we present a comprehensive dataset of online ERP-based BCIs for controlling various home appliances in diverse stimulus presentation environments. We collected online BCI data from a total of 84 subjects among whom 60 subjects controlled three types of appliances (TV: 30, door lock: 15, and electric light: 15) with 4 functions per appliance, 14 subjects controlled a Bluetooth speaker with 6 functions via an LCD monitor, and 10 subjects controlled air conditioner with 4 functions via augmented reality (AR). Using the dataset, we aimed to address the issue of inter-subject variability in ERPs by employing the transfer learning in two different approaches. The first approach, "within-paradigm transfer learning," aimed to generalize the model within the same paradigm of stimulus presentation. The second approach, "cross-paradigm transfer learning," involved extending the model from a 4-class LCD environment to different paradigms. The results demonstrated that transfer learning can effectively enhance the generalizability of BCIs based on ERP across different subjects and environments.

Novel electrotactile brain-computer interface with somatosensory event-related potential based control.

A brain computer interface (BCI) allows users to control external devices using non-invasive brain recordings, such as electroencephalography (EEG). We developed and tested a novel electrotactile BCI prototype based on somatosensory event-related potentials (sERP) as control signals, paired with a tactile attention task as a control paradigm. A novel electrotactile BCI comprises commercial EEG device, an electrical stimulator and custom software for EEG recordings, electrical stimulation control, synchronization between devices, signal processing, feature extraction, selection, and classification. We tested a novel BCI control paradigm based on tactile attention on a sensation at a target stimulation location on the forearm. Tactile stimuli were electrical pulses delivered at two proximal locations on the user's forearm for stimulating branches of radial and median nerves, with equal probability of the target and distractor stimuli occurrence, unlike in any other ERP-based BCI design. We proposed a compact electrical stimulation electrodes configuration for delivering electrotactile stimuli (target and distractor) using 2 stimulation channels and 3 stimulation electrodes. We tested the feasibility of a single EEG channel BCI control, to determine pseudo-online BCI performance, in ten healthy subjects. For optimizing the BCI performance we compared the results for two classifiers, sERP averaging approaches, and novel dedicated feature extraction/selection methods via cross-validation procedures. We achieved a single EEG channel BCI classification accuracy in the range of 75.1 to 88.1% for all subjects. We have established an optimal combination of: single trial averaging to obtain sERP, feature extraction/selection methods and classification approach. The obtained results demonstrate that a novel electrotactile BCI paradigm with equal probability of attended (target) and unattended (distractor) stimuli and proximal stimulation sites is feasible. This method may be used to drive restorative BCIs for sensory retraining in stroke or brain injury, or assistive BCIs for communication in severely disabled users.

Case-Based and Quantum Classification for ERP-Based Brain-Computer Interfaces.

Low transfer rates are a major bottleneck for brain-computer interfaces based on electroencephalography (EEG). This problem has led to the development of more robust and accurate classifiers. In this study, we investigated the performance of variational quantum, quantum-enhanced support vector, and hypergraph case-based reasoning classifiers in the binary classification of EEG data from a P300 experiment. On the one hand, quantum classification is a promising technology to reduce computational time and improve learning outcomes. On the other hand, case-based reasoning has an excellent potential to simplify the preprocessing steps of EEG analysis. We found that the balanced training (prediction) accuracy of each of these three classifiers was 56.95 (51.83), 83.17 (50.25), and 71.10% (52.04%), respectively. In addition, case-based reasoning performed significantly lower with a simplified (49.78%) preprocessing pipeline. These results demonstrated that all classifiers were able to learn from the data and that quantum classification of EEG data was implementable; however, more research is required to enable a greater prediction accuracy because none of the classifiers were able to generalize from the data. This could be achieved by improving the configuration of the quantum classifiers (e.g., increasing the number of shots) and increasing the number of trials for hypergraph case-based reasoning classifiers through transfer learning.

A new benchmark dataset for P300 ERP-based BCI applications

Because of its non-invasive nature, one of the most commonly used event-related potentials in brain-computer interface (BCI) system designs is the P300 electroencephalogram (EEG) signal. The fact that the P300 response can easily be stimulated and measured is particularly important for participants with severe motor disabilities. In order to train and test P300-based BCI speller systems in more realistic high-speed settings, there is a pressing need for a large and challenging benchmark dataset. Various datasets already exist in the literature but most of them are not publicly available, and they either have a limited number of participants or utilize relatively long stimulus duration (SD) and inter-stimulus intervals (ISI). They are also typically based on a 36 target (6×6) character matrix. The use of long ISI, in particular, not only reduces the speed and the information transfer rates (ITRs) but also oversimplifies the P300 detection. This leaves a limited challenge to state-of-the-art machine learning and signal processing algorithms. In fact, near-perfect P300 classification accuracies are reported with the existing datasets. Therefore, one certainly needs a large-scale dataset with challenging settings to fully exploit the recent advancements in algorithm design (machine learning and signal processing) and achieve high-performance speller results. To this end, in this article we introduce a new freely- and publicly-accessible P300 dataset obtained using 32-channel EEG, in the hope that it will lead to new research findings and eventually more efficient BCI designs. The introduced dataset comprises 18 participants performing a 40-target (5×8) cued-spelling task, with reduced SD (66.6 ms) and ISI (33.3 ms) for fast spelling. We have also processed, analyzed, and character-classified the introduced dataset and we presented the accuracy and ITR results as a benchmark. The introduced dataset and the codes of our experiments are publicly accessible at https://data.mendeley.com/datasets/vyczny2r4w.

A survey on encoding and decoding technology of non-invasive brain-computer interface

A survey on encoding and decoding technology of non-invasive brain-computer interface

Improving Cross-State and Cross-Subject Visual ERP-Based BCI With Temporal Modeling and Adversarial Training.

Brain-computer interface (BCI) is a useful device for people without relying on peripheral nerves and muscles. However, the performance of the event-related potential (ERP)-based BCI declines when applying it to real environments, especially in cross-state and cross-subject conditions. Here we employ temporal modeling and adversarial training to improve the visual ERP-based BCI under different mental workload states and to alleviate the problems above. The rationality of our method is that the ERP-based BCI is based on electroencephalography (EEG) signals recorded from the scalp's surface, continuously changing with time and somewhat stochastic. In this paper, we propose a hierarchical recurrent network to encode all ERP signals in each repetition at the same time and model them with a temporal manner to predict which visual event elicited an ERP. The hierarchical architecture is a simple yet effective method for organizing recurrent layers in a deep structure to model long sequence signals. Taking a cue from recent advances in adversarial training, we further applied dynamic adversarial perturbations to create adversarial examples to enhance the model performance. We conduct our experiments on one published visual ERP-based BCI task with 15 subjects and 3 different auditory workload states. The results indicate that our hierarchical method can effectively model the long sequence EEG raw data, outperform the baselines on most conditions, including cross-state and cross-subject conditions. Finally, we show how deep learning-based methods with limited EEG data can improve ERP-based BCI with adversarial training. Our code is available at https://github.com/aispeech-lab/VisBCI.

Adaptive Sequence-Based Stimulus Selection in an ERP-Based Brain-Computer Interface by Thompson Sampling in a Multi-Armed Bandit Problem.

A Brain-Computer Interface (BCI) is a device that interprets brain activity to help people with disabilities communicate. The P300 ERP-based BCI speller displays a series of events on the screen and searches the elicited electroencephalogram (EEG) data for target P300 event-related potential (ERP) responses among a series of non-target events. The Checkerboard (CB) paradigm is a common stimulus presentation paradigm. Although a few studies have proposed data-driven methods for stimulus selection, they suffer from intractable decision rules, large computation complexity, or error propagation for participants who perform poorly under the static paradigm. In addition, none of the methods have been applied to the CB paradigm directly. In this work, we propose a sequence-based adaptive stimulus selection method using Thompson Sampling in the multi-bandit problem with multiple actions. During each sequence, the algorithm selects a random subset of stimuli with fixed size, aiming to identify all target stimuli and to improve the spelling speed by reducing the number of unnecessary non-target stimuli. We compute "clean" stimulus-specific rewards from raw classifier scores via the Bayes rule. We perform extensive simulation studies to compare our algorithm to the static CB paradigm. We show the robustness of our algorithm by considering the constraints of practical use. For scenarios where simulated data resemble the real data the most, the spelling efficiency of our algorithm increases by more than 70%, compared to the static CB paradigm.

Speech stream segregation to control an ERP-based auditory BCI

Objective. The use of natural sounds in auditory brain–computer interfaces (BCI) has been shown to improve classification results and usability. Some auditory BCIs are based on stream segregation, in which the subjects must attend one audio stream and ignore the other(s); these streams include some kind of stimuli to be detected. In this work we focus on event-related potentials (ERP) and study whether providing intelligible content to each audio stream could help the users to better concentrate on the desired stream and so to better attend the target stimuli and to ignore the non-target ones. Approach. In addition to a control condition, two experimental conditions, based on the selective attention and the cocktail party effect, were tested using two simultaneous and spatialized audio streams: (a) the condition A2 consisted of an overlap of auditory stimuli (single syllables) on a background consisting of natural speech for each stream, (b) in condition A3, brief alterations of the natural flow of each speech were used as stimuli. Main results. The two experimental proposals improved the results of the control condition (single words as stimuli without a speech background) both in a cross validation analysis of the calibration part and in the online test. The analysis of the ERP responses also presented better discriminability for the two proposals in comparison to the control condition. The results of subjective questionnaires support the better usability of the first experimental condition. Significance. The use of natural speech as background improves the stream segregation in an ERP-based auditory BCI (with significant results in the performance metrics, the ERP waveforms, and in the preference parameter in subjective questionnaires). Future work in the field of ERP-based stream segregation should study the use of natural speech in combination with easily perceived but not distracting stimuli.

Semi-Supervised Learning for Auditory Event-Related Potential-Based Brain–Computer Interface

A brain-computer interface (BCI) is a communication tool that analyzes neural activity and relays the translated commands to carry out actions. In recent years, semi-supervised learning (SSL) has attracted attention for visual event-related potential (ERP)-based BCIs and motor-imagery BCIs as an effective technique that can adapt to the variations in patterns among subjects and trials. The applications of the SSL techniques are expected to improve the performance of auditory ERP-based BCIs as well. However, there is no conclusive evidence supporting the positive effect of SSL techniques on auditory ERP-based BCIs. If the positive effect could be verified, it will be helpful for the BCI community. In this study, we assessed the effects of SSL techniques on two public auditory BCI datasets-AMUSE and PASS2D-using the following machine learning algorithms: step-wise linear discriminant analysis, shrinkage linear discriminant analysis, spatial temporal discriminant analysis, and least-squares support vector machine. These backbone classifiers were firstly trained by labeled data and incrementally updated by unlabeled data in every trial of testing data based on SSL approach. Although a few data of the datasets were negatively affected, most data were apparently improved by SSL in all cases. The overall accuracy was logarithmically increased with every additional unlabeled data. This study supports the positive effect of SSL techniques and encourages future researchers to apply them to auditory ERP-based BCIs.

EEG-Inception: A Novel Deep Convolutional Neural Network for Assistive ERP-Based Brain-Computer Interfaces.

In recent years, deep-learning models gained attention for electroencephalography (EEG) classification tasks due to their excellent performance and ability to extract complex features from raw data. In particular, convolutional neural networks (CNN) showed adequate results in brain-computer interfaces (BCI) based on different control signals, including event-related potentials (ERP). In this study, we propose a novel CNN, called EEG-Inception, that improves the accuracy and calibration time of assistive ERP-based BCIs. To the best of our knowledge, EEG-Inception is the first model to integrate Inception modules for ERP detection, which combined efficiently with other structures in a light architecture, improved the performance of our approach. The model was validated in a population of 73 subjects, of which 31 present motor disabilities. Results show that EEG-Inception outperforms 5 previous approaches, yielding significant improvements for command decoding accuracy up to 16.0%, 10.7%, 7.2%, 5.7% and 5.1% in comparison to rLDA, xDAWN + Riemannian geometry, CNN-BLSTM, DeepConvNet and EEGNet, respectively. Moreover, EEG-Inception requires very few calibration trials to achieve state-of-the-art performances taking advantage of a novel training strategy that combines cross-subject transfer learning and fine-tuning to increase the feasibility of this approach for practical use in assistive applications.

Improving the Cross-Subject Performance of the ERP-Based Brain-Computer Interface Using Rapid Serial Visual Presentation and Correlation Analysis Rank.

The brain–computer interface (BCI) is a system that is designed to provide communication channels to anyone through a computer. Initially, it was suggested to help the disabled, but actually had been proposed a wider range of applications. However, the cross-subject recognition in BCI systems is difficult to break apart from the individual specific characteristics, unsteady characteristics, and environmental specific characteristics, which also makes it difficult to develop highly reliable and highly stable BCI systems. Rapid serial visual presentation (RSVP) is one of the most recent spellers with a clean, unified background and a single stimulus, which may evoke event-related potential (ERP) patterns with less individual difference. In order to build a BCI system that allows new users to use it directly without calibration or with less calibration time, RSVP was employed as evoked paradigm, then correlation analysis rank (CAR) algorithm was proposed to improve the cross-individual classification and simultaneously use as less training data as possible. Fifty-eight subjects took part in the experiments. The flash stimulation time is 200 ms, and the off time is 100 ms. The P300 component was locked to the target representation by time. The results showed that RSVP could evoke more similar ERP patterns among subjects compared with matrix paradigm. Then, the included angle cosine was calculated and counted for averaged ERP waveform between each two subjects. The average matching number of all subjects was 6 for the matrix paradigm, while for the RSVP paradigm, the average matching number range was 20 when the threshold value was set to 0.5, more than three times as much larger, quantificationally indicating that ERP waveforms evoked by the RSVP paradigm produced smaller individual differences, and it is more favorable for cross-subject classification. Information transfer rates (ITR) were also calculated for RSVP and matrix paradigms, and the RSVP paradigm got the average ITR of 43.18 bits/min, which was 13% higher than the matrix paradigm. Then, the receiver operating characteristic (ROC) curve value was computed and compared using the proposed CAR algorithm and traditional random selection. The results showed that the proposed CAR got significantly better performance than the traditional random selection and got the best AUC value of 0.8, while the traditional random selection only achieved 0.65. These encouraging results suggest that with proper evoked paradigm and classification methods, it is feasible to get stable performance across subjects for ERP-based BCI. Thus, our findings provide a new approach to improve BCI performances.

An idle state-detecting method based on transient visual evoked potentials for an asynchronous ERP-based BCI

BackgroundAn asynchronous brain-computer interface (BCI) allows subject to freely switch between the working state and the idle state, improving the subject’s comfort. However, using only the event-related potential (ERP) to detect these two states is difficult because of the small amplitude of the ERP. MethodOur previous study finds that an odd-ball paradigm could evoke transient visual evoked potentials (TSVEPs) simultaneously with ERPs. This study adopts the TSVEP and the ERP to detect the idle state in the design of an asynchronous TSVEP-ERP-based BCI (T-E BCI). The T-E BCI extracts time and frequency features from brain signals and uses a novel probability-based fisher linear discriminant analysis (P-FLDA) to combine the classification results of the ERP and the TSVEP. ResultTen subjects perform visual speller and video watching experiments, and their brain signals are measured under the working and idle states. The main results show that the T-E BCI achieves a higher accuracy than the ERP-based BCI when judging the subject’s intentions and the two states. The P-FLDA performs better than the FLDA in combining the classification results. ConclusionsThe study demonstrates that adding the TSVEP can substantially reduce the number of wrongly detected trials. The T-E BCI provides a new way of designing an asynchronous BCI without adding any additional visual stimuli, which makes the BCI more practical.

Asynchronous Control of ERP-Based BCI Spellers Using Steady-State Visual Evoked Potentials Elicited by Peripheral Stimuli.

Brain-computer interface (BCI) spellers based on event related potentials (ERPs) are intrinsically synchronous systems. Therefore, selections are constantly made, even when users are not paying attention to the stimuli. This poses a major limitation in real-life applications, in which an asynchronous control is required. The aim of this study is to design, develop and test a novel method to discriminate whether the user is controlling the system (i.e., control state) or is engaged in other task (i.e., non-control state). To achieve such an asynchronous control, our method detects the steady-state visual evoked potentials (SSVEPs) elicited by peripheral stimuli of ERP-based spellers. A characterization experiment was conducted with 5 subjects to investigate general aspects of this phenomenon. Then, the proposed method was validated with 15 subjects in offline and online sessions. Results show that the proposed method provides a reliable asynchronous control, achieving an average accuracy of 95.5% for control state detection during the online sessions. Furthermore, our approach is independent of the ERP classification stage, and to the best of our knowledge, is the first procedure that does not need to extend the duration of the calibration sessions to acquire non-control observations.

Regularized Group Sparse Discriminant Analysis for P300-Based Brain-Computer Interface.

Event-related potentials (ERPs) especially P300 are popular effective features for brain-computer interface (BCI) systems based on electroencephalography (EEG). Traditional ERP-based BCI systems may perform poorly for small training samples, i.e. the undersampling problem. In this study, the ERP classification problem was investigated, in particular, the ERP classification in the high-dimensional setting with the number of features larger than the number of samples was studied. A flexible group sparse discriminative analysis algorithm based on Moreau-Yosida regularization was proposed for alleviating the undersampling problem. An optimization problem with the group sparse criterion was presented, and the optimal solution was proposed by using the regularized optimal scoring method. During the alternating iteration procedure, the feature selection and classification were performed simultaneously. Two P300-based BCI datasets were used to evaluate our proposed new method and compare it with existing standard methods. The experimental results indicated that the features extracted via our proposed method are efficient and provide an overall better P300 classification accuracy compared with several state-of-the-art methods.

Investigation of Visual Stimulus With Various Colors and the Layout for the Oddball Paradigm in Evoked Related Potential-Based Brain-Computer Interface.

Objective: Stimulus visual patterns, such as size, content, color, luminosity, and interval, play key roles for brain–computer interface (BCI) performance. However, the three primary colors to be intercompared as a single variable or factor on the same platform are poorly studied. In this work, we configured the visual stimulus patterns with red, green, and blue operating on a newly designed layout of the flash pattern of BCI to study the waveforms and performance of the evoked related potential (ERP).Approach: Twelve subjects participated in our experiment, and each subject was required to finish three different color sub-experiments. Four blocks of the interface were presented along the edge of the screen, and the other four were assembled in the center, aiming to investigate the problem of adjacency distraction. Repeated-measures ANOVA and Bonferroni correction were applied for statistical analysis.Main results: The averaged online accuracy was 98.44% for the red paradigm, higher than 92.71% for the green paradigm, and 93.23% for the blue paradigm. Furthermore, significant differences in online accuracy (p < 0.05) and information transfer rate (p < 0.05) were found between the red and green paradigms.Significance: The red stimulus paradigm yielded the best performance. The proposed design of ERP-based BCI was practical and effective for many potential applications.

Neural mechanisms of training an auditory event-related potential task in a brain-computer interface context.

Effective use of brain-computer interfaces (BCIs) typically requires training. Improved understanding of the neural mechanisms underlying BCI training will facilitate optimisation of BCIs. The current study examined the neural mechanisms related to training for electroencephalography (EEG)-based communication with an auditory event-related potential (ERP) BCI. Neural mechanisms of training in 10 healthy volunteers were assessed with functional magnetic resonance imaging (fMRI) during an auditory ERP-based BCI task before (t1) and after (t5) three ERP-BCI training sessions outside the fMRI scanner (t2, t3, and t4). Attended stimuli were contrasted with ignored stimuli in the first-level fMRI data analysis (t1 and t5); the training effect was verified using the EEG data (t2-t4); and brain activation was contrasted before and after training in the second-level fMRI data analysis (t1 vs. t5). Training increased the communication speed from 2.9 bits/min (t2) to 4 bits/min (t4). Strong activation was found in the putamen, supplementary motor area (SMA), and superior temporal gyrus (STG) associated with attention to the stimuli. Training led to decreased activation in the superior frontal gyrus and stronger haemodynamic rebound in the STG and supramarginal gyrus. The neural mechanisms of ERP-BCI training indicate improved stimulus perception and reduced mental workload. The ERP task used in the current study showed overlapping activations with a motor imagery based BCI task from a previous study on the neural mechanisms of BCI training in the SMA and putamen. This suggests commonalities between the neural mechanisms of training for both BCI paradigms.

Performance improvement of ERP-based brain-computer interface via varied geometric patterns.

Recently, many studies have been focusing on optimizing the stimulus of an event-related potential (ERP)-based brain-computer interface (BCI). However, little is known about the effectiveness when increasing the stimulus unpredictability. We investigated a new stimulus type of varied geometric pattern where both complexity and unpredictability of the stimulus are increased. The proposed and classical paradigms were compared in within-subject experiments with 16 healthy participants. Results showed that the BCI performance was significantly improved for the proposed paradigm, with an average online written symbol rate increasing by 138% comparing with that of the classical paradigm. Amplitudes of primary ERP components, such as N1, P2a, P2b, N2, were also found to be significantly enhanced with the proposed paradigm. In this paper, a novel ERP BCI paradigm with a new stimulus type of varied geometric pattern is proposed. By jointly increasing the complexity and unpredictability of the stimulus, the performance of an ERP BCI could be considerably improved.

Learning from label proportions in brain-computer interfaces: Online unsupervised learning with guarantees.

ObjectiveUsing traditional approaches, a brain-computer interface (BCI) requires the collection of calibration data for new subjects prior to online use. Calibration time can be reduced or eliminated e.g., by subject-to-subject transfer of a pre-trained classifier or unsupervised adaptive classification methods which learn from scratch and adapt over time. While such heuristics work well in practice, none of them can provide theoretical guarantees. Our objective is to modify an event-related potential (ERP) paradigm to work in unison with the machine learning decoder, and thus to achieve a reliable unsupervised calibrationless decoding with a guarantee to recover the true class means.MethodWe introduce learning from label proportions (LLP) to the BCI community as a new unsupervised, and easy-to-implement classification approach for ERP-based BCIs. The LLP estimates the mean target and non-target responses based on known proportions of these two classes in different groups of the data. We present a visual ERP speller to meet the requirements of LLP. For evaluation, we ran simulations on artificially created data sets and conducted an online BCI study with 13 subjects performing a copy-spelling task.ResultsTheoretical considerations show that LLP is guaranteed to minimize the loss function similar to a corresponding supervised classifier. LLP performed well in simulations and in the online application, where 84.5% of characters were spelled correctly on average without prior calibration.SignificanceThe continuously adapting LLP classifier is the first unsupervised decoder for ERP BCIs guaranteed to find the optimal decoder. This makes it an ideal solution to avoid tedious calibration sessions. Additionally, LLP works on complementary principles compared to existing unsupervised methods, opening the door for their further enhancement when combined with LLP.

Improving zero-training brain-computer interfaces by mixing model estimators

Objective. Brain-computer interfaces (BCI) based on event-related potentials (ERP) incorporate a decoder to classify recorded brain signals and subsequently select a control signal that drives a computer application. Standard supervised BCI decoders require a tedious calibration procedure prior to every session. Several unsupervised classification methods have been proposed that tune the decoder during actual use and as such omit this calibration. Each of these methods has its own strengths and weaknesses. Our aim is to improve overall accuracy of ERP-based BCIs without calibration. Approach. We consider two approaches for unsupervised classification of ERP signals. Learning from label proportions (LLP) was recently shown to be guaranteed to converge to a supervised decoder when enough data is available. In contrast, the formerly proposed expectation maximization (EM) based decoding for ERP–BCI does not have this guarantee. However, while this decoder has high variance due to random initialization of its parameters, it obtains a higher accuracy faster than LLP when the initialization is good. We introduce a method to optimally combine these two unsupervised decoding methods, letting one method’s strengths compensate for the weaknesses of the other and vice versa. The new method is compared to the aforementioned methods in a resimulation of an experiment with a visual speller. Main results. Analysis of the experimental results shows that the new method exceeds the performance of the previous unsupervised classification approaches in terms of ERP classification accuracy and symbol selection accuracy during the spelling experiment. Furthermore, the method shows less dependency on random initialization of model parameters and is consequently more reliable. Significance. Improving the accuracy and subsequent reliability of calibrationless BCIs makes these systems more appealing for frequent use.

Generalized sparse discriminant analysis for event-related potential classification

A brain–computer interface (BCI) is a system which provides direct communication between the mind of a person and the outside world by using only brain activity (EEG). The event-related potential (ERP)-based BCI problem consists of a binary pattern recognition. Linear discriminant analysis (LDA) is widely used to solve this type of classification problems, but it fails when the number of features is large relative to the number of observations. In this work we propose a penalized version of the sparse discriminant analysis (SDA), called generalized sparse discriminant analysis (GSDA), for binary classification. This method inherits both the discriminative feature selection and classification properties of SDA and it also improves SDA performance through the addition of Kullback–Leibler class discrepancy information. The GSDA method is designed to automatically select the optimal regularization parameters. Numerical experiments with two real ERP-EEG datasets show that, on one hand, GSDA outperforms standard SDA in the sense of classification performance, sparsity and required computing time, and, on the other hand, it also yields better overall performances, compared to well-known ERP classification algorithms, for single-trial ERP classification when insufficient training samples are available. Hence, GSDA constitute a potential useful method for reducing the calibration times in ERP-based BCI systems.