Spatial filtering pipeline evaluation of cortically coupled computer vision system for rapid serial visual presentation

Abstract

Rapid Serial Visual Presentation (RSVP) is a paradigm that supports the application of cortically coupled computer vision to rapid image search. In RSVP, images are presented to participants in a rapid serial sequence which can evoke Event-related Potentials (ERPs) detectable in their Electroencephalogram (EEG). The contemporary approach to this problem involves supervised spatial filtering techniques which are applied for the purposes of enhancing the discriminative information in the EEG data. In this paper we make two primary contributions to that field: 1) We propose a novel spatial filtering method which we call the Multiple Time Window LDA Beamformer (MTWLB) method; 2) we provide a comprehensive comparison of nine spatial filtering pipelines using three spatial filtering schemes namely, MTWLB, xDAWN, Common Spatial Pattern (CSP) and three linear classification methods Linear Discriminant Analysis (LDA), Bayesian Linear Regression (BLR) and Logistic Regression (LR). Three pipelines without spatial filtering are used as baseline comparison. The Area Under Curve (AUC) is used as an evaluation metric in this paper. The results reveal that MTWLB and xDAWN spatial filtering techniques enhance the classification performance of the pipeline but CSP does not. The results also support the conclusion that LR can be effective for RSVP based BCI if discriminative features are available.

Use of Neural Signals to Evaluate the Quality of Generative Adversarial Network Performance in Facial Image Generation

“…where S i ∈ R 1×T is one trial reconstructed source signal. The LDA beamformer method can be applied to different time regions to reconstruct different individualized spatial profiles for ERP components present in that time frame (Wang et al, 2018b). In this study, we apply the LDA beamformer between 400 ms and 600 ms in order to best extract the P300.…”

“…Fvalues from an ANOVA test were computed for each channel across four categories. Topography is created at the optimal P300 time index for each participants which is demonstrated in (Wang et al, 2018b). Figure 7 summarizes the details from Table 3.…”

“…The reconstructed averaged P300 amplitude is used as the basis for our novel metric for evaluating GAN outputs. To address latency of the P300 which varies across participants, this work (Wang et al, 2018b) has successfully demonstrated the use of LDA beamformer to search for the optimal P300 time index in an RSVP experiment. We select the maximum value in the 200 ms time window which is centered at the optimal time index to represent the reconstructed single trial P300 amplitude and then average these across the trials to get the reconstructed averaged P300 amplitude.…”

BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks

“…Unfortunately, electroencephalogram (EEG) data are difficult to collect over long periods, and the data do not compensate for the CIP. Lee and Wang considered class imbalance as a main bottleneck factor contributing to the poor performance of RSVP classification, and data augmentation methods can help lessen the effects of CIP [2], [5], [6]. Data augmentation mainly enlarges an existing dataset by generating and transforming original sample data, which is a promising approach in computer vision.…”

Synthetic-Neuroscore: Using a neuro-AI interface for evaluating generative adversarial networks

“…By applying advanced statistical and machine learning techniques to noninvasive EEG, better source localization and reconstruction becomes possible. Our previous work [33,34] demonstrated the effectiveness of using spatial filtering approaches for reconstructing P300 source ERP signals. Remaining low SNR issues can be further remedied by averaging EEG trials.…”