Sequential feature selection

The following parameters have been added to SEQUENTIALFS*:

'groups': A cell vector where each cell contains the numbers of the columns in the initial X matrix that will be handled as one variable, i.e. selected or rejected simultaneously. The default is empty, i.e. each column is handled as separate variable

'accelerate': A number between 0 and 1. Default value is 0 (no acceleration). At every iteration, 'accelerate' is multiplied with the error score that every new feature returns. Then this quantity is compared with the previous iteration best score. Every feature that has higher weighted error compared to the best one, is either excluded from the final selection (forward selection) or included in the final selection (backward selection) without any more evaluations

*https://mathworks.com/help/stats/sequentialfs.html