However,theaggregationofdata coming from multiple subjects is challenging, since it requires accounting for largevariability inanatomy,functional topographyandstimulus response across individuals.

One of the most effectivecontinuous deep reinforcement learning algorithms is normalized advantage functions (NAF).

Unlike projection based algorithms [14,32] though, momentum does not perform well with FW.