The training set consists of input/output pairs from the target function, (x1,y1),...,(xN,yN) with yn=f(xn) for n=1,..,N

With a training set of size 100 picked according the random distribution P, p(f(x)!=g(x) over the entire range of x)=< 5% with probability > 99%

source

jl@crush.caltech.edu index

learning_algorithm

learning_problem

example_2

example_1