Data Augmentation for Brain-Computer Interfaces: Analysis on Event-Related Potentials Data

TL;DR

This paper explores data augmentation techniques for EEG-based brain-computer interfaces, demonstrating that temporal and spatial distortions can improve classification accuracy by up to 6%.

Contribution

It introduces and evaluates novel data augmentation methods specifically tailored for EEG signals, addressing the scarcity of training data in brain-computer interface applications.

Findings

Augmentation methods improve BCI performance by 1-6%.

Temporal and spatial distortions are effective for EEG data.

Augmentation enhances signal processing chain accuracy.

Abstract

On image data, data augmentation is becoming less relevant due to the large amount of available training data and regularization techniques. Common approaches are moving windows (cropping), scaling, affine distortions, random noise, and elastic deformations. For electroencephalographic data, the lack of sufficient training data is still a major issue. We suggest and evaluate different approaches to generate augmented data using temporal and spatial/rotational distortions. Our results on the perception of rare stimuli (P300 data) and movement prediction (MRCP data) show that these approaches are feasible and can significantly increase the performance of signal processing chains for brain-computer interfaces by 1% to 6%.