|
Research Institute of Intelligent Computer Systems Ternopil National Economic University |
|
2007, Vol. 6, Issue 3 |
|
Contents and abstracts
A ROBUST SPEECH RECOGNITION SYSTEM
Abderrahmane Amrouche 1), Abdelmalik Taleb-Ahmed 2),
1) Faculty of Electronics and Computer Science, USTHB, P.O. Box 32, Bab Ezzouar, Algiers, Algeria. General Regression Neural Networks (GRNN) have been applied to phoneme identification and isolated word recognition in clean speech. In this paper, the authors extended this approach to Arabic spoken word recognition in adverse conditions. In fact, noise robustness is one of the most challenging problems in Automatic Speech Recognition (ASR) and most of the existing recognition methods, which have shown to be highly efficient under noise free conditions, drastically fail drastically in noisy environments. The proposed system has been tested for Arabic digit recognition at different Signal-to-Noise Ratio (SNR) levels and under four noisy conditions: multitalker babble background, car production hall (factory), military vehicle (leopard tank) and fighter jet cockpit (buccaneer) issued from NOISEX-92 data base. The proposed scheme was successfully compared to similar recognizers based on the Multilayer Perceptron (MLP), the Elman Recurrent Neural Network (RNN) and the discrete Hidden Markov Model (HMM). The experimental results show that the use of nonparametric regression with an appropriate smoothing factor (spread) improved the generalization power of the neural network and the global performance of the speech recognizer in noisy environments.
FORMALIZATION OF SEARCH CONTEXT ON BASE OF ONTOLOGIES Anatoly Gladun 1), Julia Rogushina 2)
1) International Research and Training Centre of Information Technologies and Systems, For more relevant informational retrieval and matching of user request with metadata about informational recourses it is necessary to formulize the user knowledge about subject domain of search. We propose to use the ontologies and associated with them thesauri of the appropriate subject domains for representation of domain knowledge. The algorithms of formation and normalization of the multilinguistic thesauruses, and also methods of their comparison are given in this work.
COMPARISON BETWEEN HAAR AND DAUBECHIES WAVELET
Fatma H. Elfouly 1), Mohamed I. Mahmoud 2),
1) HIE, Alshorouk Academy, Cairo, Egypt. Recently, the Field Programmable Gate Array (FPGA) technology offers the potential of designing high performance systems at low cost. The discrete wavelet transform has gained the reputation of being a very effective signal analysis tool for many practical applications. However, due to its computation-intensive nature, current implementation of the transform falls short of meeting real-time processing requirements of most application. The objectives of this paper are implement the Haar and Daubechies wavelets using FPGA technology. In addition, the comparison between the Haar and Daubechies wavelets is investigated. The Bit Error Rat (BER) between the input audio signal and the reconstructed output signal for each wavelet is calculated. It is seen that the BER using Daubechies wavelet techniques is less than Haar wavelet. The design procedure has been explained and designed using the stat-of-art Electronic Design Automation (EDA) tools for system design on FPGA. Simulation, synthesis and implementation on the FPGA target technology has been carried out.
MULTI DIMENSIONAL DYNAMIC SCENE ANALYSIS. Panayot Iliev 1), Plamen Tzvetkov 2), George Petrov 3)
1) New Bulgarian University, Department of Telecommunications, 21 Montevideo Str, This paper presents an unconventional approach for object racking using image statistical criteria and 3D image entropy sequence analysis. The experimental results prove that the relationship between statistical characteristics of the 3D image entropy sequences and process of motion estimation is a guarantee for creating reliable and high precision target detection and tracking system. Using 2D and 3D multistage entropy functions analysis provide us a better way to reduce sequence channels for tracking moving and non moving objects. Vector based approach is used for object searching and detection inside image sequences. This way we provide necessary information for object based image sequences compression format that is much more efficient than standard MGEG video stream. Object based compression format is much more acceptable for high demanding security and military systems. SIMULATION SERVICE PROVIDING IN A DISTRIBUTED SIMULATION ENVIRONMENT Alexey Cheptsov
Donetsk National Technical University, 83000, Donetsk, vul. Artema 58, Simulation of complex dynamic systems is an interdisciplinary problem. To solve this problem the author suggests a distributed simulation environment as a new form of simulation means system organization that provides web-based distributed simulation of engineering and industrial tasks as simulation services using Discrete Event Simulation Framework on the basis of Grid infrastructure. The Unified Process of Simulation Services Development based on Unified Modeling Language is proposed.
FAILURES DETECTION METHODOLOGY IN NON RECOVERY George Popov
Technical University – Sofia, Computer Science Department,
NANOTECHNOLOGY PRINCIPLES OF OPTOELECTRONIC MODULE REALIZATION FOR Volodymyr P. Kozhemyako, Andriy A. Yarovyy, Ruslan M. Novytsky
Vinnytsia National Technical University, In given article use of the approaches, developed by the authors, is offered at designing separate structures of computing systems, which is based of optoelectronic principle of eye - processor information processing on the basis of dynamic multifunctionality, with the purpose of maintenance of development and application of nanotechnology optoelectronic high-performance elements and devices for their further integration in general structure of computing, laser and optoelectronic systems, and also systems of artificial intelligence.
ELECTRONIC NOSE FOR ANCHOVY FRESHNESS MONITORING BASED A. Amari 1), N. El Bari 2) and B. Bouchikhi 1)*
1) Sensors, Electronic & Instrumentation Group, Faculty of Sciences, Physics Department, University Moulay Ismail, An electronic nose based system, which employs an array of six inexpensive commercial gas sensors based on tin dioxide (Figaro Engineering Inc., Japan), has been used to analyse the freshness states of anchovies. Fresh anchovies were stored in a refrigerator at 4 ± 1°C over a period of 15 days. Electronic nose measurements need no sample preparation and the results indicated that the spoilage process of anchovies could be followed by using this technique. Conductance responses of volatile compounds produced during storage of anchovy were monitored and the result were analysed by multivariate analysis methods. In this paper principal component analysis (PCA) and linear discriminant analysis (LDA) were used to investigate whether the electronic nose was able to distinguishing among different freshness states (fresh, moderated and non-fresh samples). The loadings analysis was used to identify the sensors responsible for discrimination in the current pattern file. Therefore, the support vector machines (SVM) method was applied to the new subset, with only the selected sensors, to confirm that a subset of a few sensors can be chosen to explain all the variance. The results obtained prove that the electronic nose can discriminate successfully different freshness state using LDA analysis. Some sensors have the highest influence in the current pattern file for electronic nose. Support vector machine (SVM) model, applied to the new subset of sensors show the good performance. OPTIMAL DISIGN OF ENERGY SUPPLY NETS ON GRAPHS Vladimir Nikulshin 1), Viktoria von Zedtwitz 2)
1) Professor and Head of Department, Odessa National Polytechnic Univ., Ukraine, In the design and operation of energy intensive systems, the possibility of improving the system’s efficiency is very important to explore. The main way of improving efficiency is through optimisation. This paper describes the application of exergy topological models and, in particular, the graph of thermoeconomical expenditure for thermoeconomical optimal design s of circled nets for energy supply (CNES). The questions of thermoeconomical optimisation of CNES, as well as suggested modelling algorithms, are illustrated in the numerical example of the optimisation of a energy supply system for a city with seven regions of energy consumption.
ARTIFICIAL NEURAL NETWORK BASED ONLINE SENSOR Shakeb A. Khan 1), Tarikul Islam 2), Gulshan 3)
1) Department of Data Communication & Networking, College of Computer Engineering, This paper presents an artificial neural network (ANN) based generalized online method for sensor response linearization and calibration. Inverse modeling technique is used for sensor response linearization. Multilayer ANN is used for inverse modeling of sensor. The inverse model based technique automatically compensates the associated nonlinearity and estimates the measurand. The scheme is coded in MATLAB® for offline training and for online measurement and successfully implemented using NI PCI-6221 Data Acquisition (DAQ) card and LabVIEW® software. Manufacturing tolerances, environmental effects, and performance drifts due to aging bring up a need for frequent calibration, this ANN based inverse modeling technique provides greater flexibility and accuracy under such conditions.
ACOUSTIC INVARIANT APPROACH TO SPEECH SOUND ANALYSIS Maksym O. Vakulenko
Kyjiv Taras Shevchenko National University, On the basis of acoustic invariant speech analysis (AISA), the permanent spectral characteristics of the Ukrainian vowels are obtained for various ways of pronunciation including ordinary speech, whisper and changing tone. It is shown that the lowest phonemic frequencies due to vocal fold oscillations or to Helmholtz resonance are not associated with persistent sound features. It is conjectured that the only phonemic invariant is the ratio between formant frequencies, not their absolute values. This analysis is complemented by the computer sound synthesis. We show also that the acoustic invariants of the Ukrainian sound [i] are close to that of English [I]. The results obtained may be useful for specialists in the field of experimental phonetics and speech modelling. DISTRIBUTED TRUST IN ePAYMENTS SYSTEM K. P. Vidya
Department of Mathematics, Madras Christian College(Autonomous), In this paper, a secret sharing scheme that is based on the Parallel Pollard rho Attack of the Elliptic Curve Discrete Logarithm Problem (ECDLP) is proposed for hierarchical access structures that can be activated dynamically. The shares of the scheme are distributed across two levels of participants but the reconstruction of the secret takes place at level zero which is the central processor or the trusted party of the scheme. The scheme finds its application in the Payments System of Banks and facilitates the replacement of paper cheques with eCheques. It also provides an efficient method of processing the payments at the Clearing House of Banks.
THE CHOICE AND SUBSTANTIATION OF THE MATHEMATICAL S. V. Dzubin, A. V. Matsiuk, S. V. Martsenko, M. V. Pryimak
Ternopil State Technical University, Ruska str., 56, Ternopil, Ukraine, 46000 The mathematical model of electroretinogram as a linear random process which takes into account the mechanism of eye retina biopotentials forming and is suitable for solving of the measuring and diagnosing problems is developed in the work. The approach is proposed, the method is developed and the statistics are given to estimate the kernel of the linear random process as electroretinogram model. The obtained results may be used in the applied research of eye diagnosis. COMPUTER AIDED ARTICULATORY TUTOR: A SCIENTIFIC STUDY Arumugam Rathinavelu 1), Hemalatha Thiagarajan 2)
1) Dr. Mahalingam College of Engg & Technology, Pollachi, South India, This paper describes the use of Computer Aided Articulation Tutor (CAAT) to conduct phonetic training to the hearing-impaired (HI) children with inner articulators as visual cues. This Articulatory tutor was developed by using Magnetic Resonance Imaging (MRI) and computer graphics techniques. The articulators include the movement of jaw, lips, tongue and velum. Ten hearing impaired (HI) children between the ages 4 and 7 were selected and trained for 10 hours across 4 weeks on 24 words. Intelligibility of HI children was investigated to find out their performance in speech perception and articulation. The post-training results indicated that HI children improved in articulation of speech sounds placed at different contexts. This 3D visual simulation tool helped HI children in perceiving the speech information significantly.
AN ESTIMATION OF QUANTITY OF THE IDENTIFIED OBJECTS Zbishek Dombrovskiy
Ternopil National Economic University Application of tone signals from limited range low frequencies (LF) band is an effective mean for energy consumption systems (ŃS) objects identification under their control above radio channels networks. With purpose for determination of requirements to characteristics of digital tone signals processing a method of evaluation of maximum quantity of confidence-identified of ŃS objects in the network with a radio channel is offered. On the foundation of had been determined requirements for the construction of digital devices of generation and filtration of tone signals is compatible with authentication of ŃS objects the appropriate structure is offered and its confidence is explored. Graphics of dependences of probability identification of the objects vs. signal to noise ratio at the different fault probability are presented and results of their comparing analyses are given.
SELECTING KDD FEATURES AND USING RANDOM Rachid Beghdad
Faculty of sciences, 12 boulevard Bouaouina, Bejaia 06000, Algeria. The purpose of this study is to identify some higher-level KDD features, and to train the resulting set with an appropriate machine learning technique, in order to classify and predict attacks. To achieve that, a two-steps approach is proposed. Firstly, the Fisher’s ANOVA technique was used to deduce the important features. Secondly, 4 types of classification trees: ID3, C4.5, classification and regression tree (CART), and random tree (RnDT), were tested to classify and detect attacks. According to our tests, the RndT leads to the better results. That is why we will present here the classification and prediction results of this technique in details. Some of the remaining results will be used later to make comparisons. We used the KDD’99 data sets to evaluate the considered algorithms. For these evaluations, only the four attack categories’ case was considered. Our simulations show the efficiency of our approach, and show also that it is very competitive with some similar previous works. ANALYSIS OF THE MEDICAL DATA STRUCTURE USING SELFORGANISING MAPS O.V. Hodych 2), Yu.V. Nikolsky 1), V.V. Pasichnyk 1), Yu.M. Scherbyna 2)
1) National University „Lviv Polytechnics” In this article the authors discuss several approaches to high dimensional data structure analysis using Self- Organising Maps. The describe approaches utilise graphical images for the purpose of data structure interpretation. The evaluation of the discussed techniques has been performed using the real medical data from cardiology. The research, results of which are outlined in this paper, is a continuation of the earlier work related to the analysis of the same medical data. It is envisaged that results obtained in this and earlier research work will form a foundation for creation of a robust technology to be used for automation of diagnostic tasks in medicine. |