Hornik, Stinchcombe, and White used this theorem to prove an analog of it for neural networks that is often called the UAT. A probabilistic proof of this theorem applies the weak law of large numbers to a sequence of functions named for Bernstein. In its modern form, this theorem is typically stated in terms (15[1])of sub-algebras that separate (15[1])points and contain a nonzero constant. This theorem implies that the space (*) C-zero-comma-one is separable with respect to capital-L-infinity, meaning that it has a countable dense subset. For suitable functions f, this theorem gives a function p such that the maximal distance between f and p is less than epsilon. For 10 points, name this theorem that states that any (-5[1])continuous (10[1])function on a bounded interval can be arbitrarily approximated (10[1]-5[1])by a polynomial, (10[1]-5[1])developed by a German mathematician and generalized by Marshall H. Stone. ■END■ (10[4])