ОПРЕДЕЛЕНИЕ ОПТИМАЛЬНОЙ МОДЕЛИ ПРОГНОЗИРОВАНИЯ ДЛЯ ВАЛЮТНЫХ ПАР USD/EUR, USD/GBP, USD/JPYКасьян Е.А., Немирович-Данченко М

The paper provides an assessment of personal’s reliability in the personnel security system of a company. The analysis of existing solutions in various industries is carried out. The stages of assessment and levels of reliability are shown.

In this paper, we discuss the construction of fuzzy classifiers by dividing the task into the three following stages: the generation of a fuzzy rule base, the selection of relevant features, and the parameter optimization of membership functions for fuzzy rules. The structure of the fuzzy classifier is generated by forming the fuzzy rule base with use of the minimum and maximum feature values in each class. This allows us to generate the rule base with the minimum number of rules, which corresponds to the number of class labels in the dataset to be classified. Feature selection is carried out by a binary spider monkey optimization (BSMO) algorithm, which is a wrapper method. As a data preprocessing procedure, feature selection not only improves the efficiency of training algorithms but also enhances their generalization capability. In the process of feature selection, we investigate the dynamics of changes in classification accuracy, iteration by iteration, for various parameter values of the binary algorithm and analyze the effect of its parameters on its convergence rate. The parameter optimization of fuzzy rule antecedents uses another spider monkey optimization (SMO) algorithm that processes continuous numerical data. The performance of the fuzzy classifiers based on the rules and features selected by these algorithms is tested on some datasets from the KEEL repository. Comparison with two competitor algorithms on the same datasets is carried out. It is shown that fuzzy classifiers with the minimum number of rules and a significantly reduced number of features can be developed with their accuracy being statistically similar to that of the competitor classifiers.

The article proposes the application of the assessment of phrase and word intelligibility through speech recognition approach in the framework of solving the problems of speech rehabilitation after the combined treatment of oncological diseases. Speech intelligibility assessments were obtained using three speech recognition systems (Google, Yandex, Voco) and compared with expert assessments of intelligibility. Experimental results show a positive opinion about the proposed approach and they are agreed with expert assessments. Based on the processed data from rehabilitation for the Russian language, a recommendation is formulated on using the Google recognition system in the first version of the being developed product. The statistical significance of the differences in the obtained estimates of intelligibility between patient sessions and the coincidence of the sign of these differences with expert estimates and theoretical expectations are shown.

Head and neck cancer patients often have side effects that make speaking and communicating more difficult. During the speech therapy the approach of perceptual evaluation of voice quality is widely used. First of all, this approach is subjective as it depends on the listener’s perception. Secondly, the approach requires the patient to visit a hospital regularly. The present study is aimed to develop the automatic assessment of pathological speech based on convolutional neural networks to give more objective feedback of the speech quality. The structure of the neural network has been selected based on experimental results. The neural network is trained and validated on the dataset of phonemes which are represented as Mel-frequency cepstral coefficients. The neural network is tested on the syllable dataset. Recognition of the phoneme content of the syllable pronounced by a patient allows to evaluate the progress of the rehabilitation. A conclusion about the applicability of this approach and recommendations for the further improvement of its performance were made.

When developing oil reservoirs composed of carbonate rocks and those characterized by complex structure, the production well flow rate is largely determined by reservoir fracturing/ porosity rather than structure. These reservoir properties can often be reflected in a time section as significant attenuation of a seismic signal. The running time window spectral analysis has been proposed in the previous research to detect fractured zones. The calculations of seismic field diffraction due to a single pore and pore ensemble effects were made. The present research indicates that reservoir fracturing or porosity can cause qualitatively similar behavior of reflected signal amplitude spectra. Based on this finding, a rejection filter was constructed and applied to a real time section of the field in Tomsk Oblast, Prony and Fourier spectra being tested.

The article considers the effect of porous media on elastic wave field. Based on numerical modeling, diffraction pattern of the wave propagating through a single pore in carbonates has been produced. Matrix properties (calcite and dolomite) and fluid (water) are modeled based on thin core section image. The qualitative comparison with the available computational data has been performed. Provided that ensemble of pores is involved, the effect of porous medium on seismic field has been studied. For comparison with experimental data the model of porous sintered aluminum Al-6061 has been considered. The processing of numerical modeling results made it possible to estimate average velocities in the model of porous aluminum and compare them with physical modeling data. The provided estimates have indicated qualitative (single pore) and quantitative (ensemble of pores) correlation of simulation and experiment results.

The paper considers modeling technique for the medium with a network of inclined fractures and calculations for completing seismic tasks for such a medium. The network of plane-parallel fractures has controlled dip angle and fluid saturation. The time section for a model with water-saturated fractures is produced. The comparison of incident and transmitted signal spectra is made.

Numerical simulation of seismoacoustic emission (SAE) associated with fracturing in zones of shear stress concentration shows that SAE signals are polarized along the stress direction. The proposed polarization methodology for monitoring of slope stability makes use of three-component recording of the microseismic field on a slope in order to pick the signals of slope processes by filtering and polarization analysis. Slope activity is indicated by rather strong roughly horizontal polarization of the respective portion of the field in the direction of slope dip. The methodology was tested in microseismic observations on a landslide slope in the Northern Tien-Shan (Kyrgyzstan).