Oh, for a second I thought the GP was referring to nonlinear programming. Then I remembered this is an ML thread. There are some nonzero number of papers refer

Oh, for a second I thought the GP was referring to nonlinear programming. Then I remembered this is an ML thread. There are some nonzero number of papers referring to NLP (as in nonlinear programming) in ML, mostly for the purposes of constrained optimization, but I agree with your current breakdown.