In this article we propose formalization of optimal allocation problem for emergency services, to determine their abundance and professional staff under consideration of the characteristics of the territories and distribution of possible emergencies. Refs.: 12 titles.
In this article is considered the methods of digital filtering of mammograms. Were defined the features of the effect of different types of filters on methods for finding edges of pathology on mammograms. A modified method of Sobel and digital filtering were developed to improve the visualization of pathology at the mammograms. Figs.: 3. Refs.: 9 titles.
Algorithms filtering of medical images to eliminate additive, pulse and combined noise were considered. Algorithms of filtering were compared based on received quantitative estimates of image quality, visual evaluation and image's profiles. Tables of peak signal/noise ratio and the time of algorithms executing were represented. The recommendations for the selection of filters based on effectiveness and timing of their work were proposed. Figs.: 3. Tabl.: 2. Refs.: 8 titles.
Description of method of allocation of borders is given by a ranking detector and his modifications. The receipt of the special statistics for the decision-making about a border is considered. Comparison of results of allocation of borders is conducted by an detector method with other known methods. The results of application of method and method of choice of necessary thresholds are represented, that was basis of recommendations on the use algorithms of these methods. Figs. 2. Refs.: 8 titles.
Based on means of evoinformatics, solutions to global problems caused by globalization (integration, intellectualization etc.) are proposed. We analyze methodology for fragmentary simulation where on interrelations of nested systems partial models of global systems are developed. Basic conceptual principles, models and metatechnologies required to solve global problems are described. Refs.: 12 titles.
The article presents a description of the main methods of Data Mining, highlighted its advantages and disadvantages. The analysis reveals that none of the above methods are capable to extract knowledge from data. The study demonstrates the operation of Robinson's resolution method for comparing two simple sentences. It is proposed an algorithm of comparing the logic-linguistic models of textual information within the meaning. Refs.: 8 titles.
The questions of control doses to patients with X-ray computed tomography (CT) are discussed. The physical factors of CT dose index values, which form the basis for the most common method for determining radiation dose in patients with CT, are analyzed. As a result of the shortcomings of this method are analyzed and for its refinement are suggested. Figs.: 1. Tabl.: 1. Refs.: 11 titles.
Basic tasks which stand before the developers of CASS of air traffic control are considered, from the point of such providing view systems information of quality, necessary for the increase of level of automation of operations of control and decline of errors amount of controllers air courts. The index of control quality is offered, formalization of task is carried out and the method of her decision is offered. Refs.: 10 titles.
Analyzed disadvantages known Hamming neural network not capable of recognizing the images in the same minimum Hamming distance between two or more reference images. A new neural network, using the Hamming distance and read the images on the border of two or three classes. Figs.: 2. Refs.: 9 titles.
The architecture of the hybrid hierarchical neural network (HHNN) based on neural networks (NN) adaptive resonance theory ART-1, and their versions of ART-1h and ART-1s with NN Hemming. HHNN is used to describe the ontology knowledge multi-agent system control machinery manufacturer. Figs.: 2. Refs.: 10 titles.
The concepts of the normalized sub-banding correlation and sub-banding distance. A new method for segmentation of speech signals on the limits of the sounds of speech, based on the use sub-banding distance. A new method for isolation of segments of speech signals generated by the sounds of speech from the post of a periodic structure. Refs.: 8 titles.
The problem of synthesis of models based on association rules is concidered. The process of mining association rules is analyzed. Various types of association rules (negative, quantitative, generalized, temporal and fuzzy association rules) for solving data mining problems are investigated. Refs.: 15 titles.
The paper presents the material of research of features representation of educational information in the field of Law. The components of the integrated area of educational information of modern university are defined. The results of the experimental operation the training complex in the Moodle system are shown in the example of a single discipline. The guidelines for improving tests are formulated. Tasks of representation a structure of knowledge are defined in this article also. Figs.: 2. Refs.: 12 titles.
On the example of a three-layer model of human rubble study the amplitude of the reflected signal and distortion for various forms of the probe pulse. Shown that the largest amplitude is reflected probe signal in the form of Gaussian monocycle. Figs.: 7. Refs.: 18 titles.
This article discusses how to use the K-Valued neural network Adaptive resonance theory for classification K-Valued output signal of logic elements and devices of computer engineering. This allows to automate the processes of identifying situations that may cause crashes in digital devices. Figs.: 5. Refs.: 10 titles.
The paper deals with the choice of input parameters of a decision support system (DSS) technological process machined parts for aircraft engine. Are the key parameters that have practical effects on the process of machining, and shows the links between them. Refs.: 8 titles.
Considered the aspects of improving the quality of training neural networks such as multilayer perceptron, by analyzing and adjusting the parameters such as: the learning step of gradient method, capacity of the input vector parameters of regularization. Graphically shown the possible dependence of the target function of these parameters and given the means of reducing the negative influence of a particular situation. Figs: 3. Refs: 9 titles.
The mathematical model of the thremass electromechanics system is worked out taking into account the model uncertainty plant and the vector of robust control purpose is reasonable. Research of dynamic descriptions of robust control the thremass electromechanics system is executed. It is shown that the system of robust control has a subzero sensitiveness to the change of parameters of plant model. Figs.: 2. Refs.: 17 titles.
The formalized stages diagnostic-medical process at development computer decision support system in medicine. For complex estimation stage diagnostic-medical process for the reason minimization risk doctor-mistake is designed method to clusterizations diagnosis in space pharmacological action and correction threshold in diagnostic solving rule. Refs.: 8 titles.
Research heart activity sound signals (auscultation) is one of the most effective and inexpensive methods for early diagnosis of diseases of the cardiovascular system. Heart Sounds (fonokardiohrama) is a low-frequency, multicomponent, non-stationary signals, so the analysis of such signals should be doing in the frequency and time domain. Application of wavelet transform to analyze the phonocardiogram allows to increase the resolution and to obtain important information characteristics. Refs.: 15 titles
In this work the researches devoted to the decision of the task of search of several spherical surfaces in space, the points set by a three-dimensional array are continued. The approach allowing to localize in space such objects and to calculate their parameters is offered. Explicitly computing complexity of the offered algorithm is researched. Figs.: 1. Refs.: 8 titles
An example of the statistical analysis of the database, made up from 30 subjects, whose activities had being monitored by the inertial sensors: accelerometers and gyroscopes, is considered. The classification rule that allows unique identification of human activities on the signals of inertial sensors without using video was constructed by using R-package. Figs.: 2. Tabl.: 2. Refs.: 11 titles.
In this paper we consider the formalization of the initial information for the diagnosis of neurological diseases. The possibility of application of fuzzy logic and artificial neural networks. Performed structural and parametric identification of a model of health systems based on fuzzy logic to build a computer decision support system for the diagnosis of neurological diseases. Refs.: 10 titles.
The method and soft for realization of comparative analysis of mathematical models of processes and effects, which were received by methods of complete factorial and fractional factorial experiments, are developed. In result the most optimal goodness of fit of mathematical models are Fisher’s and Munn-Whitney criterion. Refs.: 11 titles.
Mathematical models of melting of reagents in liquid metall during secondary treatment are analyzed in the article. Model comparison is accomplished according to the following criteria: source data, reagents melting temperature, conditions to finish modeling process, geometrical parameters of added reagents. Mind-maps are employed in order to classify data about models. Figs.: 3. Refs.: 14 titles.
The method of prognosis of risk dynamics of professional pathology, based on application of eventual heterogeneous Markov’s chains is described. The estimation of risk depending on experience and work places on an enterprise in problem of health states space are determined by stochastic matrices. The got estimations of risk are supporting elements in prognostication of offensive of accident insured - dynamics of professional pathology at the working enterprises of chemical and pharmaceutical industry. Figs.: 2. Refs.: 10 titles.