Random subset selection in MEPX

[Mihai Oltean]

Hi there,

Now we have Random Subset Selection (RSS) in Multi Expression Programming X.

That means that you can train the algorithm on a much smaller data set. There is an edit box in the Parameters panel where the RSS size can be specified.

At each generation the RSS set is chosen randomly. 

Running time is reduced significantly if you use much smaller size for RSS. However, I do not recommend to use too small RSS sets because the algorithm can get very good results on a particular set and later will not be able to generalize on the test set.

regards,

mihai