1) It remains unclear that convergence problems occurred often in RFA/LMS method. Such problems might imply that this method should not be used for a certain type of data, resulting in the questionable use. In addition, since convergence problems occurred often in RFA/LMS method, how could the authors concluded that both LMS and RSP yielded very similar results.
2) Is it possible to add the purification procedure to this study? It might solve the convergence problems as well as improve proportions of true positives. Then, we can have a better picture regarding the performance of the three methods.