Information technology for determining groups of similar objects on the set of the studied features taking into account their changes over time has been proposed. Сomputational schemes and software for data analysis of hydrochemical monitoring has been developed.
The basic problem representation of the semantic content of text based on semantic analysis is considered. The different methods of text analysis methods for further information retrieval and document classification.
In this paper we considered methods to control the reliability of highly reliable products that work in the piecewise stationary mode. We present computational method of sequential analysis of procedure for the sample realization the continuous-time based on spline-exponential distribution with a single node, depending on the relative position of nodes of the spline.
On the basis of method of theory of markov processes, the mathematical model of conduct of concentration of hydrochemical indexes of drinking-water is developed.
The authors propose a method for estimating the parameters of the exponential distribution with a limited amount of experimental data. The advantage of the method in comparison with the method of moments based on the evaluation of arithmetic mean are shown.
There is developed method for segmentation of aggregated structures of spherical particles (emulsion drops) in dispersive formations. Method is based on information of aggregates structure and contours. Method may be used to design systems for emulsions (and other micro objects) measuring and analyzing.
A simplistic method of topic identification is proposed. The topic index is calculated as a real number, and can be used for comparing with a threshold when validating the topic identity of the content. The approach is based on the assumption of having a set of characteristic words for the estimated topic and yields a simple implementation which is suitable both for the tasks where the high precision is not required and for solving more complex problems, where it can be used in the context of preliminary topic evaluation in a multi-level topic identification process.
It was created information technology to make forecast for fuzzy time series based on fuzzy logic and genetic algorithm. This approach was used in the developed software product Fuzzy_Forecassting which was created in Eclipse on the programming language Java.
The program «SmallSample» for distributions restoration using small samples is developed.The distributions nonparametric restoration methods for small samples are compared.
A general statement of the problem of conducting non-binary subdivision from local splines is proposed. The relevant linear operators are given and the location of indications to design numerical sequences is defined.
The computing schemes of spline-distributions automated identification using probability paper are improved. They are tested on modeling data from normal, logarithmic normal, exponential and Weibull spline-distributions.
Proposed using change-point detection methods to determine areas of stationarity in time series of stock market quotations. The review of existing sequential methods of change-point detection provided.
It is proposed to use fuzzy logics for making prognosis models of financial time series. It has made comparison between Japanese candlestick theory and method based on high-order fuzzy logical relationships.
The whispered speech is considered as a form of speech, for which a problem of complex and ambiguous perceptiontake place. An algorithm for processing it in order to improve legibility and further synthesis is proposed.