Parallel memory-efficient processing of BCI data

Following after magnetic resonance imaging (MRI) and electrocortigraphy (ECoG), electroencephalography (EEG)-based research is entering the world of big data[l]. A research-quality brain-computer interface (BCI) data set can easily number in the hundreds of millions of points, making methodology of processing and classification critical. A selection of broadly applicable optimization methods implemented in R is presented that enables users to take advantage of parallelization, guaranteed call-by-reference to limit memory overhead, and scalable performance with common BCI processing tasks. As proof of concept, classification results for a P300 experiment and performance statistics are presented.