Reviews: Information-based Adaptive Stimulus Selection to Optimize Communication Efficiency in Brain-Computer Interfaces

The authors consider the problem of adaptively selecting stimulus presentations online for a BCI based on evoked potentials. They focus on the problem of presenting the best "flash groups" in a P300 spelling application. They derive a method to select stimuli that maximize mutual information between a target character and the information likely to be extracted from EEG signals (in the form of classifier scores). They derive their methods with an eye towards computational efficiency to allow their method to work online during BCI use and demonstrate their method with simulated and real, online experiments. The authors do a very nice job of motivating the need for online stimulus selection for evoked BCI work and it is clear the authors are addressing an important problem. However, I feel the paper, in it's current form, may not be appropriate for NIPS.