Abstract

Steady-State Visual Evoked Potential (SSVEP) is an important paradigm frequently used in Electroencephalography (EEG)-based brain–computer interfaces (BCI) due to its high performance and portability. However, when applying SSVEP BCItoportable and relatively cost-effective consumer-grade EEG devices, the limited number of channels compared to professional EEGequipment constrains the performance of SSVEP classifiers. Based on recent physiological research, this study proposes a set of enhancement strategies for deep learning SSVEP classifiers by leveraging symmetry. These strategies include data augmentation based on channel swapping and the addition of virtual channels as extra features. We evaluated these methods using representative models, SSVEPformer and EEGNet, under simulated sparse channel conditions. The results indicate that by applying these enhancement strategies, the performance of deep learning SSVEP classifiers is improved within short time windows.