ABSTRACT
The GMDH family of modeling algorithms used today discovers the structure (functional form) of empirical models as well as performing the traditional task of fitting model coefficients to bases of observational or postulated data. Forty years ago scientists began seeking such inductive algorithms in their quest for underlying principles governing the activity of the central nervous system. It was believed–with good reason–that a grasp of these principles would yield improvements in feedback control systems and in the design of automatic calculating machines.
