|
Home address: 14, Turgeneva Street, ap. N 1,
Simferopol, 95017,Crimea,
Ukraine.
Phone: (380652) 25-76-82
Office address: 4, Yaltinskaya Street,
Department of Mathematics,
Tavrical National University,
Simferopol,333007, Crimea,
Ukraine.
Phone: (380652) 23-03-25
Fax: (380652) 23-23-10
E-mail: donskoy@ccssu.crimea.ua
|
|
|
|
1983 Received Candidate Sc. degree in Math. Cybern. from the Computer
Center at the Russian Academy of Sciences, Moscow.
Diss. Title: Decision trees construction based learning algorithms.
1994 Received Doctor Sc. degree in Computer Science Theory from the Computer
Center at the Russian Academy of Sciences, Moscow.
Diss. Title: Discrete decision making models with incomplete information:
the synthetic approach.
1974 - 1977 Programmer, Computer Centre at Simferopol State University.
1977 - 1983 Head of Software Develop Group, Senior scientific collaborator,Simferopol
State University.
1983 - 1994 Assistant Professor, Applied MathematicDepartment, Simferopol
State University.
1994 - Chief of the Chair of ComputerScience Theory, Full professor, Simferopol
State University
Member of Russian Association ofArtificial Intelligence.Member of the
Crimean Academy of Sciences.
Member of the Society for Mathematics,Economics and Operation Research
(GMOOR),Germany.
Current research interests include Knowledge Based Systems, Pattern Recognition,
Artificial Intelligence,Discrete mathematics, Optimization with Incomplete
Data, Decision Support Systems,Expert Systems, AI+OR=IDSS.
Books
1. Verlan A.F.,Apatova N.V.,Donskoy V.I. Personal Computer'sLanguages.
Kiev,"Naukova Dumka", 1989. 236 p. (in Russian)
2. Donskoy V.I., Bashta A.I. Discrete Decision Making Modelswith Incomplete
Data. Simferopol, Tavria, 1992. 166 p.(in Russian)
Papers (more than 70)
1. Donskoy V.I. On Pattern Recognition Soft Systems Building//Programming,
Moscow, 1980, 2, p.87-89. (in Russian)
2. Donskoy V.I. Decision Trees Construction based Learningalgorithms//
J.Comp.Math.Math.Phys., Moscow, 1982, 22, 4,p.963-974. (in Russian)
3. Donskoy V.I Weakly defined linear boolean programmingproblems//J.Comp.
Math. Math.Phys., Moscow, 1988, 28, 9,p. 1379-1385. (in Russian)
4. Donskoy V.I. A statistical criterion for deciding dynamicclassification
problems//J. of Soviet Mathematics, C.B.,
New York, Vol.57, 5, 1991, p.3427-3428.
5. Donskoy V.I The Dual Expert Systems//Proceedings ofRussian Academy
of Science. Tech. Cybern., 1993, 5,
p.111-119. (in Russian)
6. Donskoy V.I. Pseudo-Boolean Optimization with aDisjunctive constraint//Comp.
Maths. Math. Phys., Pergamon,
1994, Vol.34, N2, p. 389-398
7. Donskoy V.I. Logical production systems: analysis andsynthesis//Cybernetics
and System Analysis, Kiev, 1994,4, p.11-22. (in Russian)
8. Donskoy V.I. Expert Systems: New Applicatios//Programs,Systems, Models.
Crimea Academy of Sc. Simferopol, 1995,p. 4-17. (in Russian)
9. Donskoy V.I. Intelligent Decision Making based on theCanonical Optimization
Model with a Disjunctive Constraint:Theory and Applications//Proc. of
Int. Conf. AIENG'96,11-13 Sep. 1996, Clearwater, Florida.10.Donskoy V.,
Perekhod I. Multiple Criteria Models with theLinear PseudoBoolean Functions
and Disjunctive Restrictions
//Multiple Criteria Decision Making. Proc. of the 12th Int.Conf., Hagen,
Germany. Springer, 1997, p.13-21.
|
1. Decision making with incomplete information: knowledge based
and case based optimization models (AI+OR=IDSS)
We consider optimization models based on incompleteinitial information
utilization. Cases' data base and rules' knowledge base are used for the
optimization models synthesis.The canonical form for these models representation
is a form with disjunctive constraint. Algorithms for the disjunctive
constraint synthesis using existing cases (inductive synthesis) and rules
(deductive synthesis) are described. Models with incomplete definition
of the goal function are considered as well.
2. Duality Expert Systems
Combined approach to Expert Systems development based on the inductive
and deductive methods is expounded. The deductive method uses production
rules, and inductive method uses empirical redundancies discovered by
the logical decision tree learning algorithm. Both inductive and deductive
decision rules are expressed in disjunctive normal form used for combined
decision making, comparison, and the knowledge base and/or experimental
data base edition. This so-called Dual combined approach ensures an error
probability decrease and has great advantages.
3. Decision trees synthesis for Pattern Recognition and Expert
Systems
Heuristics algoritms for the decision trees synthesis are considered.
New probability error estimators are obtained. Logical analysis of decision
trees and software realization are considered.
4. PseudoBoolean optimization with a disjunctive constraint
Algoritms for minimizing a pseudoboolean function with a constraint given
in the form of a logic equation D = 1, where D is a disjunctive normal
form, are considered. The different equivalent ways for representing the
problems of pseudoboolean optimization, their reduction to various forms,
and the complexity of their solution are investigated.
- Discrete Mathematics
- Operation Research
- Pattern Recognition
- Decision Making with Incomplete Information
- Artificial Intelligence and Expert Systems
- Introduction in Algorithms and Computational Complexity
- Programming
|
|