Novel Signal-to-Signal translation method based on StarGAN to generate artificial EEG for SSVEP-based brain-computer interfaces

作者：

Highlights：

• SSVEP-StarGAN, a novel signal-to-signal translation method for SSVEP-based BCIs, was proposed.

• Artificial SSVEP signals were successfully generated using resting EEG data.

• The classification accuracies and ITR were significantly improved using the artificial SSVEP signals.

• The artificial SSVEP signals can be used to effectively reduce the time-consuming calibration time.

摘要

•SSVEP-StarGAN, a novel signal-to-signal translation method for SSVEP-based BCIs, was proposed.•Artificial SSVEP signals were successfully generated using resting EEG data.•The classification accuracies and ITR were significantly improved using the artificial SSVEP signals.•The artificial SSVEP signals can be used to effectively reduce the time-consuming calibration time.

论文关键词：Brain–computer interface,Steady-state visual evoked potential,Generative adversarial networks,Multidomain signal-to-signal translation

论文评审过程：Received 1 February 2022, Revised 22 April 2022, Accepted 9 May 2022, Available online 13 May 2022, Version of Record 14 May 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117574