Specialization and Results of ITIM Department

The investigation strategy of the ITIM department covers a complete cycle of scientific research in the inductive modeling area:

• methodology of the modeling from data samples;
• theoryof inductive constructing models of optimal complexity;
• algorithmizationof high-performance inductive modeling tools;
• intellectualizationof tools and technologies for constructing models;
• computer experiments to evaluate the effectiveness of the developed algorithms and tools;
• solving real-world problemsof modeling and forecast;
• testing and applied implementationof the developed tools in monitoring, control and decision support systems.

Starting from 1998, when the Department has been established, there was developed:

• the GMDH-based inductive modeling theorywith the use of the method of critical variances[1] which made it possible to explain the nature of the GMDH efficiency as a method for constructing noise-immune models with minimum prediction error variance, as well as to solve the problem of optimizing the data sample partitioning into two parts [2].
• two-criteriamethod for extra-determining the model choiceusing the errors unbiasedness criterion [3] eliminating the ambiguity in choosing the optimal model.
• principles of designing and implementing high-performance sorting-out GMDH algorithmsbased on recurrent calculations[4], paralleling operations[5] and sequential selectionof informative variables [6], allowing to enhance the dimensionality of the problems being solved.
• principles of constructing hybrid architectures of iterative GMDH algorithms as a generalization of algorithmic structures of multilayered, relaxational and combinatorial types, based on which a generalized iterative algorithmGIA GMDH [7] was developed as a neural network with active neurons in the form of the COMBI algorithm for automatic adjustment of a neuron complexity.
• thegeneralized relaxational iterative algorithm GRIA GMDH based on the use of high-speed recurrent computations and matrices of normal equations, which allows solving inductive modeling problems from high-dimension data [2].
• theoretical foundations of a new class of sorting-out GMDH algorithms with the use of recurrent-and-parallel computations on cluster systems [8] as a basis for high-performance intelligent technologies of inductive modeling.
• principles of designing technologies for the intelligent modelingof complex systems based on the use of knowledge bases, inductive data analysis tools and an intelligent user interface [9]. Such technologies should have three main instrumental levels: autonomousmodeling from the available database;embeddedmodeling as part of a real-time control system; combined modeling of a complex system for identifying optimal operation modes and critical scenarios.
• theoretical principles and tools forpredicting interrelated socio-economic processes from statistical data in the class of discrete dynamic models of vector autoregression [10];
• principles of hybridizationof sorting-out GMDH and genetic algorithms, based on which a searching algorithm COMBI-GA has been constructed [11].

1. ASTRID methodology of designing GMDH-based computer technologies for building models of complex systems from statistical data for discovering regularities, identification, prediction, aimed to informational support of decision-making problems.
2. Package of software tools[4] for designing, researching and applying modeling methods, conducting experiments on testing modeling methods and their components (model classes, generators of model structures, methods for estimating parameters and models selection criteria).
3. Cross-platformsoftware package with an advanced interface[12] in the Java language for inductive modeling and forecasting of complex objects and large-scale processes on the basis of the fast-operating GMDH algorithm with sequential selection of informative and/or sifting of non-informative arguments [6].
4. The software package ASTRID-GIA [13] for inductive modeling of complex systems based on various iterative GMDH algorithms makes it possible to use the generalized algorithm GIA GMDH and all its special cases [7] in online access mode both over the Internet and in the local network.
5. Computer system ASPIS [14] for building predictive models on the basis of the high-speed generalized relaxational iteration algorithm GRIA GMDH [2]. It is implemented in C++ and allows solving big data problems.
6. The management decisions informational support system MDISS allows to solve problems of estimation, analysis and forecasting of the state of complex systems of interrelated socio-economic processes with the purpose of making reasonable decisions [15].

The tasks of modeling of the following processes were solved:

• dynamics of the quantity of microorganisms in the soil, depending on environmental factors and the dose of contamination with heavy metals [16];
• dynamics of changing interdependent indicators of the energy and investment areas in the class of vector autoregression models for a short-term forecast [4], [10];
• dependence of the sputtering (disruption) coefficient of a spacecraft surface under action of ionized gas jets on the physical properties of the surface coating [4];
• analysis of the ecological consequences of trees irrigation with treated wastewater [17].
• quantitative assessment of the impact of sea water pollution with bitumoid substances on the total number of species of benthic organisms in the Sevastopol bays [18].
• predicting the results of testing blood samples with medical products in order to determine the most effective for a particular patient [2];
• classifiers construction for the differential diagnostics of blood diseases for reducing the risks of misdiagnosis [2];
• enhancing the relevance of the text information retrieval using GIA GMDH [19].

[1] V.S. Stepashko, "Method of Critical Variances as Analytical Tool of Theory of Inductive Modeling," J. of Automation and Information Sciences, 2008, vol. 40, no. 2, pp. 4-22.

[2] A.V. Pavlov, V.S. Stepashko, N.V. Kondrashova, Effective methods of models self-organization, Kyiv: Akademperiodika, 2014, 200 p. (In Russian)

[3] A.G. Ivakhnenko, E.A. Savchenko, "Investigation of Efficiency of Additional Determination Method of the Model Selection in the Modeling Problems by Application of GMDH Algorithm", J. of Automation and Information Sciences, 2008, vol. 40, no. 2, pp. 47-58.

[4] V.S. Stepashko, S.M. Yefimenko, Ie. A. Savchenko, Computer experiment in inductive modeling, Kyiv: NaukovaDumka, 2014.- 222 p. (In Ukrainian)

[5] V. Stepashko, S. Yefimenko, "Parallel algorithms for solving combinatorial macromodelling problems," Przeglad Elektrotechniczny (Electrical Review), 2009, vol. 85, no. 4, pp. 98-99.

[6] O. Samoilenko, V. Stepashko, "A Method of Successive Elimination of Spurious Arguments for Effective Solution of the Search-Based Modelling Tasks," Proc. of the II Int. Conf. on Inductive Modelling, Sept. 2008, pp. 36-39.

[7] V. Stepashko, O. Bulgakova, "Generalized Iterative Algorithm GIA GMDH," Proc. of the 4th Int. Conf. on Inductive Modelling ICIM-2013, Kyiv, Ukraine.- Kyiv: IRTC ITS NASU, Sept. 2013, pp. 119-123.

[8] S. Yefimenko, V. Stepashko, "Intelligent Recurrent-and-Parallel Computing for Solving Inductive Modeling Problems," Proc. of 16th Int. Conf. on Computational Problems of Electrical Engineering, Lviv, Ukraine, Lviv: LNPU, Sept. 2015, pp. 236-238.

[9] V.S. Stepashko, "Conceptual fundamentals of intelligent modeling," Control Systems and Machines (USiM), 2016, no. 4, pp. 3-15. (In Russian)

[10] S.N. Yefimenko, "Construction of systems of predictive models for multidimensional interrelated processes," Control Systems and Machines (USiM), 2016, no. 4, pp. 80-86. (In Russian)

[11] V. Stepashko, O. Moroz, "Hybrid Searching GMDH-GA Algorithm for Solving Inductive Modeling Tasks," IEEE Int. Conf. on Data Stream Mining & Processing, Lviv, Ukraine, pp. 350-355, August 2016.

[12] O.A. Samoilenko, "Designing of new GMDH algorithms as basic components of a modeling subsystem," Inductive Modeling of Complex Systems, issue 3, Kyiv: IRTC ITS NASU, 2011, pp. 191-208. (In Ukrainian)

[13] O. Bulgakova, V. Zosimov, V. Stepashko, "Software package for modeling of complex systems based on iterative GMDH algorithms with the network access capability," System Research and Information Technologies, 2014, no. 1, pp. 43-55. (In Ukrainian)

[14] A. Pavlov, "Design Patterns of Automated Structure-Parametric Identification System," Proc. of 6th Intern. Workshop on Inductive Modelling, 2015, Kyiv-Zhukyn, Ukraine.- Kyiv: IRTC ITS NASU, July 2015, pp. 31-35.- ISBN 978-966-02-7648-2.

[15] V. Stepashko, O. Samoilenko, R. Voloschuk, "Informational Support of Managerial Decisions as a New Kind of Business Intelligence Systems," In: Computational Models for Business and Engineering Domains, G. Setlak, K. Markov (Eds.), Rzeszow, Poland; Sofia, Bulgaria: ITHEA, 2014, pp. 269-279.

[16] Iutynska G., Stepashko V. Mathematical modeling in the microbial monitoring of heavy metals polluted soils. - Book of Proceedings of IX ESA Congress, 4-7 September 2006, Warsaw, Poland.- Warsaw-Pulavy: Institute of Soil Science and Plant Cultivation, 2006.- Part 2.- P. 659-660.

[17] I.K. Kalavrouziotis, V.A. Vissikirsky, V.S. Stepashko, P.H. Koukoulakis, "Application of qualitative analysis techniques to the environmental modeling of plant species cultivation," Global NEST Journal, 2010, vol. 12, no. 2, pp 161-174.

[18] S.V. Alyomov, O.S. Bulgakova, V.S. Stepashko, "Modeling of the Black Sea pollution impact on the total number of benthic organisms species," Collected articles of SNUNE&I, Sevastopol, issue 3 (39), 2011, pp. 54-62. (In Ukrainian)

[19] V. Zosimov, V. Stepashko, O. Bulgakova, "Inductive building of search results ranking models to enhance the relevance of the text information retrieval," Proc. of the 26th Intern. Workshop "Database and Expert Systems Applications", Valencia, Spain, Ed. by Markus Spies at al., Los Alamitos: IEEE Computer Society, Sept., 2015, pp. 291-295.