This paper investigates a new problem for neural information processing in the age of large-scale recording. The proposed algorithm is relatively simple (which is a positive). With promising application and openly available code/data, summing up reviewers opinions, this paper could be accepted.