Abstract: When training machine-learning models for brain–computer interface (BCI) decoding, practitioners often grapple with a large hyper-parameter space, where exhaustive tuning is hampered by ...