Universal Joint Feature Extraction for P300 EEG Classification Using Multi-Task Autoencoder

Abstract

The process of recording Electroencephalography (EEG) signals is onerous and requires massive storage to store signals at an applicable frequency rate. In this work, we propose the Event-Related Potential Encoder Network (ERPENet); a multi-task autoencoder-based model, that can be applied to any ERP-related tasks. The strength of ERPENet lies in its capability to handle various kinds of ERP datasets and its robustness across multiple recording setups, enabling joint training across datasets. ERPENet incorporates Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM), in an autoencoder setup, which tries to simultaneously compress the input EEG signal and extract related P300 features into a latent vector. Here, we can infer the process for generating the latent vector as universal joint feature extraction. The network also includes a classification part for attended and unattended events classification as an auxiliary task. We experimented on six different P300 datasets. The results show that the latent vector exhibits better compression capability than the previous state-of-the-art semi-supervised autoencoder model. For attended and unattended events classification, pre-trained weights are adopted as initial weights and tested on unseen P300 datasets to evaluate the adaptability of the model, which shortens the training process as compared to using random Xavier weight initialization. At the compression rate of 6.84, the classification accuracy outperforms conventional P300 classification models: XdawnLDA, DeepConvNet, and EEGNet achieving 79.37% -88.52% classification accuracy depending on the dataset. INDEX TERMS Electroencephalography, P300, Deep learning, Pre-trained model, Spatiotemporal neural networks, multi-task autoencoder arXiv:1808.06541v2 [eess.SP]

A Practical EEG-Based Human-Machine Interface to Online Control an Upper-Limb Assist Robot

“…Nevertheless, SSVEP was not very acceptable for some applicable scenarios, such that some cases do not need to be too fast but need to be more stable and comfortable to use. A shortcoming cannot be ignored is that we did not compare our method in-depth with some recent great P300 detection methods based on deep learning (Ditthapron et al, 2019 ). However, we think our methods on the basis of classical machine learning also have decent performance and are convenient to be implemented with lower computation cost, which is easier to be used in practice for relevant developers.…”

EEG Classification of Forearm Movement Imagery Using a Hierarchical Flow Convolutional Neural Network

“…Recent BCI advances adopted deep learning techniques, which already yielded a dramatically high performance in other research fields, such as computer vision and natural language processing [33]. In particular, a few studies applied the deep networks to various BCI paradigms using EEG signals such as mental state detection [48]- [50], emotion recognition [51], [52], intention decoding using steady-state visual evoked potentials [16], P300 [53], [54], and MI [32], [34]- [36], [42]. Several studies for MI decoding using deep learning approaches focused on enhancing decoding performance for basic multi-classes (e.g., left hand, right hand, and foot) using a public dataset [35], [36].…”

Data Analytics in Steady-State Visual Evoked Potential-Based Brain–Computer Interface: A Review

“…For example, Gao et al [113] designed a convolutional neural network with long short-term memory (CNN-LSTM) architecture, which extracts the spectral, spatial as well as temporal features of SSVEPs in order to achieve the high classification performance. However, Ditthapron et al [114] stated that it is complicated and costly to collect a large number of EEG signals for training CNN-LSTM architecture, so a pre-trained model called event-related potential encoder network (ERPENet) was proposed to classify the attended and unattended event. Generally, the pre-trained model can be fine-tuned and then employed to a novel related scenario to solve insufficient data and detection accuracy problem [5].…”

“…For instance, Embrandiri et al [115] employed denoising autoencoder to pre-train the network and then the network was trained by back-propagation to maximize contrast/SNR, which proves the feasibility of pre-trained model in SSVEP detection. Therefore, the advanced ERPENet in [114] proposed for ERP/P300 classification may provide potential direction for SSVEP-based BCI systems, which can ease the pressure of store and analyze large-scale data.…”