Reviews: Learning brain regions via large-scale online structured sparse dictionary learning

Technical quality: I would rate marginally below 3 if given the option. The model definition was clear. The algorithms followed a well established framework [14], and appeared solid. However, maybe a bit undermined by its presentation, the paper did not seem to clearly demonstrate the empirical advantage of introducing the Sobolev prior in the results section. Especially, in Figure 2, it was not clear in what aspect the SSMF method was better than the two alternatives.