An improved EEGNet for single-trial EEG classification in rapid serial visual presentation task

Abstract

As a new type of brain–computer interface (BCI), the rapid serial visual presentation (RSVP) paradigm has attracted significant attention. The mechanism of RSVP is detecting the P300 component corresponding to the target image to realize fast and correct recognition. This paper proposed an improved EEGNet model to achieve good performance in offline and online data. Specifically, the data were filtered by xDAWN to enhance the signal-to-noise ratio of the electroencephalogram (EEG) signals. The focal loss function was used instead of the cross-entropy loss function to solve the classification problems of unbalanced samples. Additionally, the subject-specific data were fed to the improved EEGNet model to obtain a subject-specific model. We applied the proposed model at the BCI Controlled Robot Contest in World Robot Contest 2021 and won the second place. The average recall rate of the four participants reached 51.56% in triple classification. In the offline data benchmark dataset (64 subjects-RSVP tasks), the average recall rates of groups A and B reached 76.07% and 78.11%, respectively. We provided an alternative method to identify targets based on the RSVP paradigm.

Improving the performance of P300-based BCIs by mitigating the effects of stimuli-related evoked potentials through regularized spatial filtering

“…Lawhern et al [27] developed EEGNet, a compact CNN-based DNN that can analyze and classify brain signals from various mental activities. Additionally, there have been improved versions of EEGNet, such as the one by Zhang et al [28] for detecting single-trial P300 signals, as well as new DNN architectures like ST-CapsNet [29], which integrates spatial and temporal attentions using a capsule network for P300 detection, and a CNN-based approach by Du et al [30] that classifies single-trial P300 signals by fusing data from multiple subjects.…”

ST-CapsNet: Linking Spatial and Temporal Attention With Capsule Network for P300 Detection Improvement

“…Most recently, with graphics processing units (GPUs) becoming more powerful, deep learning has grown tremendously. Zhang et al proposed an improved EEGNet [19] that combined xDAWN saptial filtering with EEGNet [20] for the individually-calibrated rapid serial visual presentation (RSVP) task and won second place in the BCI Controlled Robot Contest at 2022 World Robot Contest [21]. Wang et al proposed denoising autoencoder neural networks to improve the symbol recognition accuracy by about 0.7% compared to ESVMs, which can automatically learn features from unlabeled data and solve the problem of local minima in neural networks due to random initialization [22].…”

Single-trial P300 classification algorithm based on centralized multi-person data fusion CNN

“…P300 classification detection is the focus of P300-BCI research, and fast and accurate recognition is crucial to improving the performance of P300-BCI (Huang et al, 2022). The P300 usually exhibits a low signal-to-noise ratio (SNR) (Zhang et al, 2022). In order to highlight its time-locked component and minimize the background noise, P300-BCI demands collecting, aggregating and averaging data from multiple trials to obtain a reliable output (Liu et al, 2018), which is time consuming and inefficient.…”

“…Ma et al (2021) proposed a capsule network-based model that improved the detection accuracy of single-trial P300, however, the calculation became complicated due to the increase in size. Zhang et al (2022) filtered the data with xDAWN to improve the signal-to-noise ratio of EEG signals, but the spatial filtering method required manual selection of significant features after feature extraction, and then classifying them. It is highly specific to particular factors; however, the algorithm is often complex and its accuracy is influenced by feature selection (Zhang et al, 2022).…”