Classiiers That Approximate Functions Classiiers That Approximate Functions

A classiier system, XCSF, is introduced in which the prediction estimation mechanism is used to learn approximations to functions. The addition of weight vectors to the classiiers allows piecewise-linear approximation, where the classiier's prediction is calculated instead of being a xed scalar. The weight vector and the classiier's condition co-adapt. Results on functions of up to six dimensions show high accuracy. The idea of calculating the prediction leads to the concept of a generalized classiier in which the payoo prediction approximates the environmental payoo function over a subspace deened by the classiier condition and an action restriction speciied in the classiier, permitting continuous-valued actions.