Double-criterion Choice of an Optimal Model in GMDH Algorithms. Yevgeniya Savchenko

Abstract. Investigation and choice of reasonable sequence of the external GMDH criteria for the optimal model selection is the purpose of the work. The inductive modelling is directed to the optimal model construction by the adjusted external criteria. The sequence search of model by orthogonal criteria for the removal of these failings is proposed. Two variants of optimal model search are considered with a different sequence of external criteria of accuracy and bias for the choice a reasonable sequence of criteria. Numerical examples show search the reasonable sequence of the external GMDH criteria.

Keywords. Inductive modelling, GMDH, criteria bias and regularity, double-criterion choice.

References.
1. Ivakhnenko A.G., Stepashko V.S. Pomekhoustoychivost' modelirovaniya (Noise - immunity of modeling): Kiev.- Naukova Dumka. - 1985. - 216 p. (in Russian).
2. Ivakhnenko A.G., Ivakhnenko G.A., Savchenko E.A. GMDH Algorithm for Optimal Model Choice by the External Error Criterion with the Extension of Definition by Model Bias and Its Applications to the Committees and Neural Networks // Pattern Recognition and Image Analysis - 2002 - 12, N4. - p.347-353.
3. Ivakhnenko A.G., Stepashko V.S., Khomovnenko M.G., Galyamin E.P. Selforganization of the dynamics models of agricultural cultures growth for crop rotation irrigated control // Automation. - 1977. - no.5. - P. 32 - 44 (in Russian).
4. Akishin B.A., Yunusov N. Sequential applying of selforganization and statistical testing of hypothesis for adequate mathematical models choice // Automation.- 1978.- no.3. - p.64-67 (in Russian).
5. Ivakhnenko A.G., Savchenko E.A., Ivakhnenko G.A. GMDH algorithm for optimal model selection by error criterion with additional determination by bias criterion and its application in committee and neural networks // Proceedings of the International Conference of Inductive Modelling 2002. - Lviv, 2002.- P. 47- 56 (in Ukranian).
6. Ivakhnenko A.G., Savchenko E.A., Ivakhnenko G.A., Nadiradze A.B., Rogov A.O. Inductive method of Choosing a Model with the Least Error and Bias for Solving Interpolation Tasks of Artificial Intelligence // Pattern Recognition and Image Analysis.-2003.-V.13, N3.-P. 452-458.