Search academic texts

Information × Registration Number 0219U101405, 0116U002094 , R & D reports Title The development of methods for identification and analysis of complex structures of objects based on Markov models popup.stage_title Head Rаspopov Viktor B., Кандидат фізико-математичних наук Registration Date 16-07-2019 Organization Research and educational center of applied informatics NASU popup.description2  The relational scheme is considered as a generative construct and the problem of constructing a scheme of minimum size, which generates a given "language" - a set of mappings of a set of objects in a set of labels, is investigated. A basic tool is proposed for solving this problem in the form of a complete system of equivalent transformations of relational schemes. It is shown that tasks of constructing a minimum DNF correspond to the problem of constructing a system of relational schemes that generates a given language. For the probabilistic prediction of health status, as an individual, and for different groups of people for the subsequent period of life, the mathematical apparatus of the Markov chains was used. A method for calculating the forecast according to the initial data was developed. On the basis of genetic databases, five hundred mutations associated with hematologic diseases were tested using the DNA code. It is established that precisely the violation of the polarity of amino acids causes a significant effect on the occurrence of diseases. A computerized approach to the recognition of a hematologic disease is proposed based on the application of computer models of Markov chains and Bayesian procedures. Algorithms based on Bayesian procedures allowed to identify the most informative features for the diagnosis of diseases. The number of medical tests to resolve the diagnosis can be reduced several times. The mathematical properties of the standard genetic code and hypothetical codes with the improved characteristics of noise immunity are investigated. Optimized codes are built. Creation of information support programs for collecting human and animal tissue cell cultures based on a specially designed database. The report consists of three sections and an appendix, the total volume is 247 pages, fig. - 30, tab. - 20, list of links. - 163 names. Key words: models and methods of forecasting, Markov chain models, classification, recognition, consistent labe Product Description popup.authors Brovarnyk Vyacheslav V. Vagis Olexandra A. Gupal Anatoliy M. Tkachov Igor I. popup.nrat_date 2020-04-02 Close