forum.alglib.net

Hi,

I am using this algorithm to do a mapping/fitting between 2 3-dimensional spaces. It's performance is very good although there are some testing points (samples not used to build the model) with some errors that I would like to reduce even further. In any case, kudos to the developer(s). I will continue playing with the radius, the number of layers and the lambda value to see if we can improve the results and I will also double check if these particular cases should be consider outliers (measurement errors?).
Now my question is as follows: provided I have found a goof fit/approximation, is there any way to get an "explicit" expression of the "function" that the rbf-ml algorithm has built?

Thanks,

Hector