Wyniki 1-8 spośród 8 dla zapytania: authorDesc:"Pavel URBANOVICH"

Mathematical model of control points placing for monitoring polluting substances emissions in the atmosphere on the basis of the NP-full coloring graph task decision

Czytaj za darmo! »

The mathematical model of the control sensors for the atmospheric air state arrangement on the basis of NP-difficult problem solution about the graph coloring is proposed. This model is differed from analogs by the fact that the calculation speed for developed model is approximately two times higher. The model can work with the maps sized to 100×100 cells, while analogs - with maps 50×50 cells. The model also differs by the presence of additional constraints for the sensors arrangement inside the cluster. Streszczenie. W niniejszym artykule został przedstawiony matematyczny model rozmieszczenia czujników, wykorzystywanych do kontrolowania stanu powietrza atmosferycznego na podstawie rozwiązania NP-trudnego problemu kolorowania grafu. Różni się on od analogicznych modeli szybkością obliczeń, która dla proponowanego modelu jest około dwa razy wyższa. Model ten wykorzystuje tablice o rozmiarach 100×100 komórek, podczas gdy analogiczne rozwiązania korzystają z tablic o rozmiarach 50×50 komórek. Dodatkowo przedstawiony model wyróżnia się obecnością dodatkowych ograniczeń rozmieszczenia czujników wewnątrz klastra. (Model matematyczny rozmieszczenia czujników, wykorzystywanych do kontrolowania stanu powietrza atmosferycznego na podstawie rozwiązania NP-trudnego problemu kolorowania grafu) Keywords: Control sensors placement, mathematical model, NP-difficult problem solution. Slowa kluczowe: : Rozmieszczenie czujników kontroli, model matematyczny, rozwiązanie problemu NP-trudnego. Introduction For carrying out of polluting substances emissions in atmosphere from stationary sources the great value has a choice of necessary points control for placing of the sensors. These sensors are supervising the maintenance of thrown out substances in atmosphere near to a sanitaryprotective zone of the enterprise. The more quantity of sensors will be established, the more exact results will be received. However, it will also lead to substantial growth of the[...]

W-cyclic method of interleaving of the data for communication systems

Czytaj za darmo! »

A modified cycling interleaver for complex data transformation (W-cyclic method) based on redundant coding is described. Algorithms of interleaving and deinterleaving are considered. A comparative analysis of the efficiency of information transformation on the basis of W-cyclic method is carried out. Streszczenie. Została opisana zmodyfikowana metoda cyklicznego przeplotu danych (metoda W-cyclic), opierająca się na kodowaniu nadmiarowym. Przeanalizowane zostały także algorytmy przeplotu i odwrotnego przeplotu. Dodatkowo została opisana porównywalna analiza efektywności przekształcenia informacji za pomocą metody W-cyclic. (Metody w-cykliczne przeplotu danych w systemach komunikacji). Keywords: interleaver, reverse interleaving, communication systems, complex transformation of information. Słowa kluczowe: przeplot, przeplot odwrotny, systemy komunikacyjne, złożona transformacja informacji. Introduction Currently communication systems require high speed transmission and transformation of information which ensures the required level of reliability. However, they are exposed to various types of interference: electromagnetic field of meteorological conditions, operating conditions, etc. This leads to errors in data transmission [1]. Research has shown that the errors are self-dependent and have a tendency to clustering. The result is that there is a loss and distortion of information [2]. To avoid errors methods based on a robust information coding are applied. The improvement of the efficiency of coding schemes to neutralize the errors of higher multiplicity can be achieved through the sharing of correcting code and an interleaver, known as a complex transformation of information [3]. The role of the interleaver is to reduce the correlation between adjacent symbols by reversing characters. This allows converting grouping errors to a single one. Efficiency of the interleaver is determined by the the minimum distance of erroneous bits [...]

Tuning a CMU Sphinx-III Speech Recognition System for Polish Language DOI:10.12915/pe.2014.04.042

Czytaj za darmo! »

In this paper, authors describe parameters which may be tuned to obtain the best performance and accuracy for a large vocabulary continuous speech recognition task. Behavior of certain parameters should be similar regardless of the language speech recognition. However, some parameters will have a different impact on the accuracy of the Polish speech recognition as compared to the English speech recognition. Streszczenie. W niniejszym artykule autorzy opisują paramenty, które mogą być dostosowywane, w celu uzyskania większej wydajności i dokładności w zadaniach rozpoznawania mowy ciągłej. Zachowania pewnych parametrów powinny być podobne bez względu na używany język. Jednakże niektóre parametry będą miały inny wpływ na dokładność rozpoznawania mowy polskiej w porównaniu do zadań rozpoznawania mowy angielskiej (Strojenie systemu rozpoznawania mowy CMU Sphinx-III dla języka polskiego). Keywords: speech recognition, CMU Sphinx, polish language. Słowa kluczowe: rozpoznawanie mowy, CMU Sphinx, język polski. doi:10.12915/pe.2014.04.42 Introduction CMU Sphinx-III is one of the most popular speech recognition systems [1]. It works very well in continuous speech recognition tasks with a lot of words, regardless of speaker. However, to achieve satisfactory results, system must be trained on the appropriate set of utterances with the reference transcription. In addition to the training, important aspect is the appropriate tuning parameters of the decoder (speech recognition system), in this way to achieve pretty good results in a reasonable time. Sphinx system was frequently tested and analyzed in relation to the English language [2-4] and other languages [5, 6]. With regard to the Polish language publications can also be found using the system Sphinx [7]. However, there is lack of publications, that have made the analysis of the most important parameters responsible for the accuracy of speech recognition and performance time for the Polish [...]

The use of pitch in Large-Vocabulary Continuous Speech Recognition System DOI:10.15199/48.2016.08.21

Czytaj za darmo! »

Abstract. In this article the authors normalize the speech signal based on the publicly available AN4 database. The authors added to the algorithm of calculating the MFCC coefficients, the normalization procedure, that uses pitch of the voice. As demonstrated by empirical tests authors were able to improve speech recognition accuracy rate of about 20%. Streszczenie. W niniejszym artykule autorzy normalizują sygnał mowy wykorzystując publicznie dostępną bazę danych AN4. Autorzy dodali do algorytmu obliczania współczynników MFCC, procedurę normalizacji, wykorzystującą wysokość tonu głosu. Jak wynika z przeprowadzonych testów, autorzy uzyskali poprawę dokładności rozpoznawania mowy o około 20% (Wykorzystanie wysokości tonu głosu w systemach rozpoznawania mowy ciągłej z dużą ilością słów). Keywords: speech recognition, CMU Sphinx, pitch, speech normalization. Słowa kluczowe: rozpoznawanie mowy, CMU Sphinx, wysokość głosu, normalizacja sygnału mowy. Introduction The fundamental frequency (F0) plays a very important role in generating speech signal [1]. It has long been known that, formant frequencies generated by women and men are different from each other. Women have higher formant frequencies than men [2], which may be explained by the longer vocal tracts of men [3]. The important source of the inter-speaker variations is the vocal tract length (VTL). Therefore, vocal tract length normalization (VTLN) technique was described in many publications [4 - 6]. However, information about VTL is rarely used by speech recognition systems. Most often they use MFCC coefficients for each frame calculated in the same way. This means that continuous speech recognition systems do not take into account the changes in the spectrum. These changes appear in the utterance of the same phoneme spoken by various speakers. One of the most commonly used technique for the normalization is bilinear transform (BLT) [7-9]. It was presented in [10] using BTL in[...]

Using Full Covariance Matrix for CMU Sphinx-III Speech Recognition System DOI:10.15199/48.2018.07.25

Czytaj za darmo! »

CMU Sphinx-III is one of the most popular speech recognition systems [1]. It works very well in continuous speech recognition tasks with a lot of words, regardless of speaker. However, to achieve satisfactory results, system must be trained on the appropriate set of utterances with the reference transcription. The whole process of speech recognition by decoder starts with acquisition of utterance. Then, the extraction process is performed of the most desirable features (from the point of view of speech recognition system). Decoder analyzes these features using acoustic model, language model and vocabulary. Block diagram is shown in Fig.1. Fig.1. Block diagram of speech recognition system CMU Sphinx-III is a system that uses statistical methods. Namely, this system is based on a hidden Markov model (HMM). It is now the dominant solution for the most recently designed speech recognition systems. If we have a good learning set (of appropriate size and of appropriate quality) the system gives very good results (word error rate is approximately 15%). To obtain very good results training set size should take into account the following recommendations:  1 hour of recording for command and control for single speaker  5 hours of recordings of 200 speakers for command and control for many speakers  10 hours of recordings for a single speaker dictation  50 hours of recordings of 200 speakers for many speakers dictation We briefly describe the signal processing front end of the Sphinx III speech recognition system. The front end transforms a speech waveform into a set of features to be used for recognition, specifically, mel-frequency cepstral coefficients (MFCC). The front end processing performed by the Sphinx-III:  pre-emphasis  windowing (Hamming window)  power spectrum  mel spectrum  mel cepstrum. Sphinx III uses the following features:  Sample[...]

The appearance of conflict when using the chaos function for calculating the hash code

Czytaj za darmo! »

The paper considers the hash function built on artificial neural network (ANN). The data about the process of synchronization of an ANN, obtained by experiment, are presented. The fact, that the obtained vector of weight coefficients for the networks after the synchronization is different for each new session, is determined. Streszczenie. W artykule opisane zostały funkcje skrótu zbudowane w oparciu o architekturę sztucznej sieci neuronowej (SSN). Przedstawiono wyniki eksperymentalne procesu synchronizacji z SSN. Udowodniono, że uzyskane współczynniki wektora wag sieci po synchronizacji są inne dla każdej nowej sesji. (Wygląd konfliktu podczas wykorzystywania funkcji chaosu do obliczeń skrótowych). Keyword: neural network, hash function. Słowa kluczowe: sieci neuronowe, funkcja skrótu. Introduction The technology of artificial neural networks (ANN) usage for modeling cryptographic communication systems is a new, promising direction in the field of information security. There are three basic learning algorithms of ANN: with a teacher, without a teacher, with reinforcements. ANN interaction model of "learning without a teacher" has properties such as peer education, self-study, and stochastic behavior, low sensitivity to noise, inaccuracies (data corruption, weight coefficients, errors in the program). This can be used for solving of crypto conversion tasks in systems with a public key, key distribution, messages hashing and generating random[...]

Modeling and software implementation of fibrous waste disposal processes DOI:10.15199/48.2016.08.09

Czytaj za darmo! »

The article describes the development of service-oriented computer system for situational management of industrial enterprise waste management and decision support systems in a production environment. The software is based on mathematical model, which allows to solve optimization problems in the subject area and to analyze the results. Streszczenie. W artykule przedstawiono opracowany zorientowany na usługi komputerowy system do zarządzania sytuacyjnego przemysłowymi kompleksowymi systemami gospodarowania odpadami i systemami wspomagania decyzji w środowisku produkcyjnym. Oprogramowanie jest oparte na modelu matematycznym, co pozwala na rozwiązanie problemów optymalizacyjnych w tej dziedzinie oraz analizę wyników. (Modelowanie i wdrażanie oprogramowania procesów unieszkodliwiania odpadów włóknistych). Keywords: mathematical modeling, software, fibrous waste disposal processes. Slowa kluczowe: madelowanie matematyczne, oprogramowanie, proces unieszkodliwiania odpadów włóknistych. Introduction Today the problem of fibrous waste disposal (waste paper, short fibers from the shearing of artificial fur, leather waste in the form of chrome shavings and, glass, asbestos, carbon and metal fibers, etc.) is vital. All processes of the fibrous waste recycling are significantly dependent on factors of external environment (weather, wind conditions, and, above all, from the salt composition of water used). There is no only one optimal technological regime of utilization. It is necessary to develop a response to environmental changes, correcting values of the process control effects. An analysis of the collected data reveals hidden patterns and gives the opportunity to calculate, but not to guess the necessary corrections. There are some examples of effective use of data mining in the processing of the results interpretation for the fixed long-term[...]

A method of syntactic text steganography based on modification of the document-container aprosh DOI:10.15199/48.2018.06.15

Czytaj za darmo! »

The development of information technology has led to the fact that now they put the information on a par with the generally accepted material values. Access to it with the advent of global computer networks has become incredibly simple. Simplicity and speed of such access significantly increased both the threat of data security breach, as well as the threat of unauthorized access (without author's permission) to information. The main objects of copyright relating to information technologies, including publishing technologies, are: paper and electronic versions of various text or other documents, databases, computer programs. This means that Internet objects also refer to intellectual property. The logic of regulating the Internet itself and the relations connected with the use of its resources and capabilities should be in the plane of both national and international law. Thus, the problem of protecting information and protecting copyright for text documents is becoming increasingly important. A method of solving of this problem on the basis of text steganography is investigated and analyzed in some articles, for example [1]. The specifics of algorithmic and software implementation of the steganographic methods that are based to protect of digital documents against unauthorized use are analyzed in [2]. Contrary to cryptography which purpose is hiding data by encrypting them, the purpose of steganography is to hide the fact of the transfer of confidential messages. It is because of steganography system of protection is achieved the greatest degree of resistance to intentional attacks to destroy or to identify hidden information. The steganographic system (stegosystem or steganosystem) - a set of tools and techniques that are used to form a secret channel of information transfer [3-5]. The steganosystem forms the channel, that carries the filled container. This channel is considered to be exposed to the influence from the [...]

 Strona 1