Ашық рұқсат Ашық рұқсат  Рұқсат жабық Рұқсат берілді  Рұқсат жабық Тек жазылушылар үшін

Том 27, № 3 (2018)

Article

Spectral Characteristics of a Finite 2D Ising Model

Karandashev I., Kryzhanovsky B., Malsagov M.

Аннотация

The paper gives the results of a numerical simulation of a two-dimensional Ising model built on finite lattices of dimension L = 50, 100, …, 500. Approximate analytical formulae for the spectral energy density are offered. Derived from Onsager’s solution with consideration of the finite size of the system, the formulae agree well with the simulation results.

Optical Memory and Neural Networks. 2018;27(3):147-151
pages 147-151 views

A New Type of a Wavelet Neural Network

Efitorov A., Dolenko S.

Аннотация

Wavelet transformation uses a special basis widely known for its unique properties, the most important of which are its compactness and multiresolution (wavelet functions are produced from the mother wavelet by transition and dilation). Wavelet neural networks (WNN) use wavelet functions to decompose the approximated function. However, for a standard wavelet basis with fixed transition and dilation coefficients, the decomposition may be not optimal. If no inverse transformation is needed, the values of transition and dilation coefficients may be determined during network training, and the windows corresponding to various wavelet functions may overlap. In this study, we suggest a new type of a WNN—Adaptive Window WNN (AWWNN), designed primarily for signal processing, in which window positions and wavelet levels are determined with a special iterative procedure. Two modifications of this new type of WNN are tested against linear model and multi-layer perceptron on Mackey-Glass benchmark problem.

Optical Memory and Neural Networks. 2018;27(3):152-160
pages 152-160 views

Algorithm of Definition of Mutual Arrangement of L1–L5 Vertebrae on X-ray Images

Kurochka K., Panarin K.

Аннотация

When diagnosing osteochondrosis it is important to determine geometrical parameters and mutual arrangement of vertebrae. We propose an algorithm for partial automatization of the localization of the vertebrae on X-ray images of lumbar spine and their following parametrization. The algorithm is a combination of different approaches. To localize positions of the vertebrae on the image, we use the method of a sliding window with fixed size and a convolution neural network as a classificator. The following processing of the localized segments of the images with vertebrae consists of removing noise, restoration, correction, and parametrization, which we perform using the library of computer vision OpenCV.

Optical Memory and Neural Networks. 2018;27(3):161-169
pages 161-169 views

A New Neural Network Classifier Based on Atanassov’s Intuitionistic Fuzzy Set Theory

Giveki D., Rastegar H., Karami M.

Аннотация

This paper proposes a new framework for training radial basis function neural networks (RBFNN). Determination of the centers of the Gaussian functions in the hidden layer of RBF neural network highly affects the performance of the network. This paper presents a novel radial basis function using fuzzy C-means clustering algorithm based on Atanassov’s intuitionistic fuzzy set (A-IFS) theory. The A-IFS theory takes into account another uncertainty parameter which is the hesitation degree that arises while defining the membership function and therefore, the cluster centers converge to more desirable locations than the cluster centers obtained using traditional fuzzy C-means algorithm. Furthermore, we make use of a new objective function obtained by Atanassov’s intuitionistic fuzzy entropy. This objective function is incorporated in the traditional fuzzy C-means clustering algorithm to maximize the good points in the class. The proposed method is used to improve the functionality of the Optimum Steepest Descent (OSD) learning algorithm. Adjusting RBF units in the network with great accuracy will result in better performance in fewer train iterations, which is essential when fast retraining of the network is needed, especially in the real-time systems. We compare the proposed Atanassov’s intuitionistic radial basis function neural network (A-IRBFNN) with fuzzy C-mean radial basis function neural network (FCMRBFNN) while both methods use OSD learning algorithm. Furthermore, the proposed A-IRBFNN is compared with other powerful fuzzy-based radial basis function neural network. Experimental results on Proben1 dataset and demonstrate the superiority of the proposed A-IRBFNN.

Optical Memory and Neural Networks. 2018;27(3):170-182
pages 170-182 views

Context Interpolation of Multidimensional Digital Signals in Problem of Compression

Gashnikov M.

Аннотация

We analyzed algorithms of interpolation of multidimensional digital signals based on a context modeling. The proposed interpolation algorithms are adaptive due to using different parameters that interpolate functions for each reading of a digital signal. We optimized these parameters of the interpolating functions over decimate versions of the digital signal and then use them for less decimated versions of the same signal. As a result we obtained interpolation algorithms that are hierarchical and that allowed us to use them in the framework of a hierarchical compression method of multidimensional signals. We implemented our context interpolators as a program that was a part of the hierarchical compression method. Our computer simulations showed an increase of efficiency of the hierarchical compression method on account of application of the proposed interpolators based on the context modeling.

Optical Memory and Neural Networks. 2018;27(3):183-190
pages 183-190 views

The Study of the Surface Distribution of the Electrooptical Properties of the Medium by Reflected Light

Kniazkov A.

Аннотация

The results of the experimental study of the surface distribution of the electrooptic coefficients of crystals LiNbO3, SBN and PLZT-ceramics are described. The induced birefringence in medium was created by an alternating field. The harmonic analysis of reflected laser light was used. The light was reflected from the front face of the transverse cell of the electro-optical medium. The transverse cells were made in the form of planar and three-dimensional flat capacitors. The homogeneity of the surface distribution of electro-optical coefficients of the medium is estimated.

Optical Memory and Neural Networks. 2018;27(3):191-195
pages 191-195 views

Holograms Form Factor and the Recording Laser Radiation Mode Structure

Privalov V., Shemanin V., Shoydin S.

Аннотация

The lasers application in the information systems have to be preceded by the development of the requirements to the laser radiation parameters. It is clear, that the lasers radiation coherence growth by means of which the holograms were registered leads the holograms quality improvement.

Optical Memory and Neural Networks. 2018;27(3):196-202
pages 196-202 views

Application of Artificial Neural Networks in the Process of Catalytic Cracking

Muravyova E., Timerbaev R.

Аннотация

Industrial production is one of the promising areas of application of artificial neural networks (ANN). There is a tangible trend towards manufacturing modules with a high level of automation in this area, which requires an increase in the number of intelligent self-regulating and self-adjusting objects. However, industrial processes are characterized by a large variety of dynamically interacting parameters, which complicate the creation of adequate analytical models. Modern industrial production is constantly becoming more complicated. This slows down the introduction of new technological solutions. In addition, in some cases, successful analytical mathematical models show an inadequacy due to a lack of computing power. In this regard, there is an increasing interest in alternative approaches to modeling industrial processes using ANN, which provide the possibility to create models that operate in real time with small errors that can be trained in the process of use. The advantages of neural networks make their use attractive for solving problems such as: forecasting, planning, designing of automated control systems, quality management, manipulator and robotics management, process safety management: fault detection and emergency situations prevention, process management: optimization of industrial process regimes, monitoring and visualization of supervisory reports. Neural networks can be useful in industrial production, for example, when creating an enterprise risk management model, planning a production cycle. Modeling and optimization of production is characterized by high complexity, a large number of variables and constants, defined not for all possible systems. Traditional analytical models can often be built only with considerable simplification, and they mostly have evaluative nature. While the ANN is trained on the basis of data from a real or numerical experiment.

Optical Memory and Neural Networks. 2018;27(3):203-208
pages 203-208 views

Communication with Quantum Limited Subspace

Yu F.

Аннотация

One important aspect of our universe is that one cannot get something from nothing; there is always a price to pay. In this article we show that every bit of information is limited by a quantum unit. Since we are communicating within a temporal subspace, this unit can be equivalently described as a quantum limited subspace (QLS), as imposed by the Heisenberg Principle. We show that communication can be exploited within and outside the QLS. The size of a QLS is determined by carrier signal bandwidth; that is narrower the bandwidth the larger the size of the QLS. By manipulating the size of a QLS, more efficient information transmission strategies can be developed. Examples for inside and outside QLS communication are given. Extension to relativistic communication has also demonstrated. We remark that, a new era of communication is anticipated to immerge and it will change our way in communicating, observation and computing, we used to use, forever!

Optical Memory and Neural Networks. 2018;27(3):209-217
pages 209-217 views