Given a sequence of observable variables {(x1,y1),…,(xn,yn)}{(x_1, y_1), ldots, (x_n, y_n)}{(x1,y1),…,(xn,yn)}, the conformal prediction methodology estimates a confidence set for yn+1y_{n+1}yn+1 given xn+1x_{n+1}xn+1 that’s legitimate for any finite pattern measurement by merely assuming that the joint distribution of the info is permutation invariant. Though enticing, computing such a set is computationally infeasible in most regression issues…
See paper particulars