Our results suggest that EEGNet is robust enough to learn a wide variety of interpretable features over a range of BCI tasks. Our models can be found at: https://github.com/vlawhern/arl-eegmodels.

"Kernel Flux: a whole-head 432-magnetometer optically-pumped magnetoencephalography (OP-MEG) system for brain activity imaging during natural human experiences,"

No abstract

Humans can fluidly adapt their interest in complex environments in ways that machines cannot. Here, we lay the groundwork for a real-world system that passively monitors and merges neural correlates of visual interest across team members via Collaborative Brain Computer Interface (cBCI). When group interest is detected and co-registered in time and space, it can be used to model the task relevance of items in a dynamic, natural environment. Previous work in cBCIs focuses on static stimuli, stimulus-or response-locked analyses, and often within-subject and experiment model training. The contributions of this work are twofold. First, we test the utility of cBCI on a scenario that more closely resembles natural conditions, where subjects visually scanned a video for target items in a virtual environment. Second, we use an experiment-agnostic deep learning model to account for the real-world use case where no training set exists that exactly matches the end-users' task and circumstances. With our approach we show improved performance as the number of subjects in the cBCI ensemble grows, and the potential to reconstruct ground-truth target occurrence in an otherwise noisy and complex environment.