Similar ideas to what is described here are achieved by P Splines and Shape Constrained Additive Models.
Originally shared by @slwu89 at How to embed expert knowledge /constraints in ML training? - #7 by slwu89.
Similar ideas to what is described here are achieved by P Splines and Shape Constrained Additive Models.
Originally shared by @slwu89 at How to embed expert knowledge /constraints in ML training? - #7 by slwu89.
@RoyiAvital, thank you for the references.
I’ve also found the GeneralizedSmoothingSplines.jl package that allows setting spline constraints (upper/lower bounds and/or monotonicity).
The :increasing
constraint produces the following result: