Novel neural network training methods (commonly referred to as deep learning) have emerged in recent years. Using a combination of unsupervised pre-training and subsequent fine-tuning, deep neural networks have become one of the most reliable classification methods. Since deep neural networks are especially powerful for high-dimensional and non-linear feature vectors, electroencephalography (EEG) and event-related potentials (ERPs) are one of the promising applications. Furthermore, to the authors' best knowledge, there are very few papers that study deep neural networks for EEG/ERP data. The aim of the experiments subsequently presented was to verify if deep learning-based models can also perform well for single trial P300 classification with possible application to P300-based brain-computer interfaces. The P300 data used were recorded in the EEG/ERP laboratory at the Department of Computer Science and Engineering, University of West Bohemia, and are publicly available. Stacked autoencoders (SAEs) were implemented and compared with some of the currently most reliable state-of-the-art methods, such as LDA and multi-layer perceptron (MLP). The parameters of stacked autoencoders were optimized empirically. The layers were inserted one by one and at the end, the last layer was replaced by a supervised softmax classifier. Subsequently, fine-tuning using backpropagation was performed. The architecture of the neural network was 209-130-100-50-20-2. The classifiers were trained on a dataset merged from four subjects and subsequently tested on different 11 subjects without further training. The trained SAE achieved 69.2% accuracy that was higher (p < 0.01) than the accuracy of MLP (64.9%) and LDA (65.9%). The recall of 58.8% was slightly higher when compared with MLP (56.2%) and LDA (58.4%). Therefore, SAEs could be preferable to other state-of-the-art classifiers for high-dimensional event-related potential feature vectors.

• Klíčová slova
• P300, brain-computer interfaces, deep learning, event-related potentials, machine learning, stacked autoencoders,
• Publikační typ
• časopisecké články MeSH

Guess the number is a simple P300-based brain-computer interface experiment. Its aim is to ask the measured participant to pick a number between 1 and 9. Then, he or she is exposed to corresponding visual stimuli and experimenters try to guess the number thought while they are observing event-related potential waveforms on-line. 250 school-age children participated in the experiments that were carried out in elementary and secondary schools in the Czech Republic. Electroencephalographic data from three EEG channels (Fz, Cz, Pz) and stimuli markers were stored. Additional metadata about the participants were collected (gender, age, laterality, the number thought by the participant, the guess of the experimenters, and various interesting additional information). Consequently, we offer the largest publicly available odd-ball paradigm collection of datasets to neuroscientific and brain-computer interface community.

• MeSH
• dítě MeSH
• elektroencefalografie MeSH
• lidé MeSH
• mladiství MeSH
• rozhraní mozek-počítač * MeSH
• Check Tag
• dítě MeSH
• lidé MeSH
• mladiství MeSH
• mužské pohlaví MeSH
• ženské pohlaví MeSH
• Publikační typ
• časopisecké články MeSH
• práce podpořená grantem MeSH
• Geografické názvy
• Česká republika MeSH

BACKGROUND: The event-related potentials technique is widely used in cognitive neuroscience research. The P300 waveform has been explored in many research articles because of its wide applications, such as lie detection or brain-computer interfaces (BCI). However, very few datasets are publicly available. Therefore, most researchers use only their private datasets for their analysis. This leads to minimally comparable results, particularly in brain-computer research interfaces. Here we present electroencephalography/event-related potentials (EEG/ERP) data. The data were obtained from 20 healthy subjects and was acquired using an odd-ball hardware stimulator. The visual stimulation was based on a three-stimulus paradigm and included target, non-target and distracter stimuli. The data and collected metadata are shared in the EEG/ERP Portal. FINDINGS: The paper also describes the process and validation results of the presented data. The data were validated using two different methods. The first method evaluated the data by measuring the percentage of artifacts. The second method tested if the expectation of the experimental results was fulfilled (i.e., if the target trials contained the P300 component). The validation proved that most datasets were suitable for subsequent analysis. CONCLUSIONS: The presented datasets together with their metadata provide researchers with an opportunity to study the P300 component from different perspectives. Furthermore, they can be used for BCI research.

• Klíčová slova
• Event-related potentials, LED, P300, Three-stimulus paradigm, Visual stimulation,
• Publikační typ
• časopisecké články MeSH
• dataset MeSH