How to select an optimal features subset using fscmrmr ?

Hi, I'm trying to use mrmr in order to select features representing neural connection (80 object and 3570 features). Once I obtained the rank, how can I select an optimal subset of them without a brute force approach (trying to classify using 10/20/30... features per time)?

[Ive J]

You simply can't, fscmrmr ranks predictors/features. One better way would be to use a couple of other feature selection approaches and decide based on their union (although it poses other complexitieis, but should give a good starting point): https://www.mathworks.com/help/stats/feature-selection.html

Also, I don't think feature selection methods (even approaches like fscnca) guarantee an accuracy improvment of your classification/regression ML problem.