|
Research Institute of Intelligent Computer Systems Ternopil National Economic University |
|
2005, Vol. 5, Issue 2 |
|
Contents and abstracts
MODULAR AND SELF-ORGANIZING CONNECTIONIST SYSTEMS: TOWARD HIGHER LEVEL INTELLIGENT FUNCTIONS Kurosh Madani
Image, Signal and Intelligent Systems Laboratory (LISSI / EA 3956), Senart Institute of Technology, Recent advances in “neurobiology” allowed highlighting some of key mechanisms of animal intelligence. Among them one can emphasizes brain’s “modular” structure and its “self-organizing” capabilities. The main goal of this paper is to show how these primary supplies could be exploited and combined in the frame of “soft-computing” issued techniques in order to design intelligent artificial systems emerging higher level intelligent behavior than conventional Artificial Neural Networks (ANN) based structures. AN AGENT APPROACH FOR PROVIDING SECURITY IN DISTRIBUTED SYSTEMS Nataliya Kussul, Andriy Shelestov, Serhiy Skakun
Space Research Institute NASU-NSAU, 40 Glushkov Ave, Kiev, 03680, Ukraine, In this paper an agent approach for providing security in distributed systems such as computer networks, Grid systems is presented. This approach envisages on-line and off-line monitoring in order to analyze users’ activity. The monitoring is done with the use of intelligent methods, namely neural networks. SMART SILICON SENSORS BASED ON VERTICAL HALL EFFECT DEVICES Konstantin Dimitrov, Chavdar Roumenin
Institute of control and system research at the Bulgarian academy of sciences,
1113 Sofia, BULGARIA; P.O. Box 79
Future integrated systems will benefit significantly from the progress in batch manufactured silicon sensors
and signal processing techniques. Silicon technologies make possible to produce sensing microdevices combining
maximal sensitivity, high accuracy and minimal design complexity. Smart sensors on the base of Vertical vector Hall
effect devices offer a number of advantages including reducing mass, volume and power consumption; greater
redundancy of system functions and simpler architecture. In view of these characteristics, it can be expected that such
smart sensors will be used extensively wide if an adequate solution is found to reduce the design cost and simplify the
electrical interface. Consequently, cost effective microsystems including vector magnetic sensors, circuits and
eventually actuators can be fabricated. NTEGRATION OF ARTIFICIAL NEURAL NETWORKS FOR IDENTIFICATION OF COMPUTER SYSTEMS STATES Oksana Pomorova
Department of System Programming Khmelnitsky National University, The main principles of methodology of intellectualization computer systems diagnosing process are presented in paper.Offered the information model, method and means of computer systems states clusterization provide an opportunity of diagnosing on the basis of the incomplete diagnostic information. For identification of computer systems states are used the union of neural nets experts which are constructed with use of artificial neural networks architecture ART2 and SOM. REAL TIME DATA ACQUISITION SYSTEM FOR THE ECP-EPP PARALLEL PORT BASED ON PIC16F877 MICROCONTROLLER John A. Kalomiros
Technical and Educational Institute of Serres, Greece, The design of a simple and low cost 10-bit data acquisition system is presented which makes use of the peripherals of a PIC16F877 microcontroller, interfacing with a personal computer using the extended capabilities of the parallel port. The system is integrated with a visual programming tool based on LabVIEW data acquisition software, which provides design flexibility and real time signal processing capabilities. An optimum assembly code for the PIC microcontroller allows for a free-running mean sampling rate of 100KSps on a Pentium PC running Windows XP OS. This system can be an example of a low cost integrated approach for data acquisition that includes a microcontroller, a personal computer and visual measurement software. The system can be the basis of a A/D interface for many measurement applications and can also be seen as an educational paradigm in itself. An effective and fast DAC solution is also presented in full integration with the microcontroller and the computer parallel port. NEW METHOD FOR CONSTRUCTION OF OPTIMAL SCALAR QUANTIZERS FOR LAPLACIAN SOURCE Zoran Peric 1), Jelena Nikolic 1), Dragoljub Pokrajac 2)
1) Faculty of Electronic Engineering, University of Nis, 18000 Nis, Aleksandra Medvedeva 14, Serbia, AN MPI-BASED FRAMEWORK FOR PARALLEL PROCESSING OF INTEGRATED CIRCUITS LAYOUT IMAGES Aleksej Otwagin, Alexander Doudkin
United Institute of Informatics Problem, 6, Surganova st., Minsk, 220012, Belarus, We consider basic algorithms and processing technologies for integrated circuit layout images. The images represented as a set of frames can regard as a dataflow and the processing are perfectly suited for parallel implementation. We propose a framework architecture for designing parallel systems of image dataflow processing. The framework uses the algorithm of a virtual associative network for increasing processing speed and system throughput during runtime. REVIEW OF INDUSTRIAL APPLICATIONS OF COMPUTATIONAL INTELLIGENCE Colin M. Frayn
CERCIA, School of Computer Science, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK Bringing cutting-edge natural computation research into industry is a challenging task, with numerous obstacles to overcome. In this paper I describe a few selected projects that we have developed at the Centre of Excellence for Research in Computational Intelligence and Applications (CERCIA) and I discuss the promising future of Computational Intelligence techniques in industry. MATHEMATICAL MODEL OF PHOTOPLETYSMIC SIGNAL AS THE BASE FOR INFORMATIONAL PARAMETERS IDENTIFICATION Borys Martchenko, Bogdana Mlynko, Mykhaylo Fryz
Departament of computer sciences of Ternopil State Ivan Pul'uj Technical University, Taking into account the biophysical genesis of researched signal and its rhythmic nature the new mathematical model of photopletysmic signal as a linear periodical random process is developed; characteristic functions analysis of signal, experimental informational parameters estimating are possible with this model. ERROR ANALYSIS OF RICHARDSON’S EXTRAPOLATIONS Nikolay Petrov
Trakian University, Stara Zagora, Yambol, Bulgaria We propose estimators of a round off error contained in an approximation for Richardson’s extrapolation scheme under finite digit arithmetic. We also propose a stopping criterion, based on consideration of the round off error, for Richardson’s extrapolation scheme with respect to risk technical systems (automobile and railway transport, aircrafts, marine and river transport, chemical installations, munitions, information society suffering by terrorism). Usually the error of an approximation is evaluated by a truncation error. However, we can accurately estimate the behavior of this error utilizing both truncation and round off errors under finite digit arithmetic. HEURISTIC TECHNIQUES FOR HANDWRITTEN SIGNATURE CLASSIFICATION Marcin Adamski, Khalid Saeed
Faculty of Computer Science, Bialystok Technical University, Wiejska 45A, 15-351 Bialystok, Poland New theoretical and experimental techniques for offline classification of handwritten signatures are introduced in this paper. The proposed algorithms are mainly based on boundary tracing technique for extracting characteristic features. Outer and inner boundaries of the signature image are treated separately. The upper and lower parts of the boundaries are extracted to form two sequences of points. Three algorithms for calculating feature vectors are applied based on y coordinate, distances between consecutive points and from polar coordinates system. Experiments on classification of the resulted vectors were carried out by means of Dynamic Time Warping algorithm using window and slope constraints. A brief comparison between the authors' work and other known signature techniques is also discussed in the paper. COMPARATIVE ANALYSIS OF NEURAL NETWORKS AND STATISTICAL APPROACHES TO REMOTE SENSING IMAGE CLASSIFICATION Nataliya Kussul 1), Serhiy Skakun 1), Olga Kussul 2)
1) Space Research Institute NASU-NSAU, Glushkov Ave 40, Kyiv, 03650, Ukraine, inform@ikd.kiev.ua This paper examines different approaches to remote sensing images classification. Included in the study are statistical approach, in particular Gaussian maximum likelihood classifier, and two different neural networks paradigms: multilayer perceptron trained with EDBD algorithm, and ARTMAP neural network. These classification methods are compared on data acquired from Landsat-7 satellite. Experimental results showed that to achieve better performance of classifiers modular neural networks and committee machines should be applied. SOFT COMPUTING AS A SOLUTION TO TIME/COST DISTRIBUTOR Nabil M. Hewahi Computer Science Department, Islamic University of Gaza, Gaza, Palestine, nhewahi@iugaza.edu In this paper we present a theoretical model based on soft computing to distribute the time/cost among the industry/machine sensors or effectors based on the type of the application. One of the most unstudied significant work is to recognize which sensor in an industry for example has higher priority than others. This is important to know which sensor to be checked first and within time limits of the system response. The problem of such systems is their variant environmental situations. Based on these varied situations, the priority of the importance of each sensor might change from time to another. Due to this uncertainty and lack of some information, soft computing is considered to be one of the plausible solutions. The presented idea is based on initially training of the system and continuously exploiting the system experience of the degree of importance of the sensors. The proposed system has three main stages, the first stage is concerned with training the system to obtain the necessary system time to respond, the necessary time allocated to recognize which sensors to check (or which has higher priority), and the initial importance value for each sensor, which indicates the initial judgment about the sensor importance. The second stage is to use the system experience about the importance of the sensor using fuzzy logic to decide the final values of each sensor 's importance. Based on the output of the second stage and the output of the first stage, the system distributes the time/cost among the sensors (some sensors with lower priority might be neglected). The main idea of the proposed work is based on neurofuzzy. SIMULATION MODELING OF NEURAL CONTROL SYSTEM FOR SECTION OF MINE VENTILATION NETWORK Iryna Turchenko, Volodymyr Kochan, Anatoly Sachenko
Research Institute of Intelligent Computer Systems Static and dynamic simulation models of a section of a mine ventilation network in order to research a sequential neural control scheme of mine airflow are developed in this paper. The techniques of neural network training set creation for both simulation models, a structure of neural network and its training algorithm are described. The simulation modeling results using static and dynamic models have showed good potential capabilities of neural control approach. |