About
175
Publications
31,947
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
905
Citations
Introduction
Additional affiliations
January 2012 - October 2015
March 1995 - November 2014
January 1995 - present
Publications
Publications (175)
This paper thoroughly describes the use of unsupervised adaptive resonance theory ART2 neural network for the purposes of image color recognition of x-ray images and images taken by nuclear magnetic resonance. In order to train the network, the pixel values of RGB colors are regarded as learning vectors with three values, one for red, one for green...
In the current paper is presented an application of the InterCriteria Decision Making (ICDM) approach to data extracted from the Bulgarian University Ranking System [13] for the years 2012–2014. The aim is to analyze the correlations between groups of indicators against which specialty "Communication and Computer Technology" is measured.
This paper represents a prototype embedded system, intended for measuring, gathering and transmitting the obtained data of the temperature of the skin and the ambient air. The system is designed to collect data in relatively short intervals, for fairly long periods of time. The skin temperature measurements are performed using an infrared temperatu...
A data base of 140 diesel fuels having cetane number in the range 10 ? 70 points and the physical properties: density at 15?C, and distillation characteristics according to ASTM D-86 T10%, T50%, and T90% was used to develop new procedures for prediction of diesel cetane number by application of the least squares method (LSM) using MAPLE software, a...
A chemometric approach using artificial neural network for classification of biodiesels was developed. It is based on artificial neural network in its classic form Multilayer Perceptron. Gas chromatography (GC) and Gas Chromatography - mass spectrometry (GC-MS) were used for quantitative and qualitative analysis of biodiesels, produced from differe...
In this paper the fast learning algorithm of unsupervised adaptive resonance theory ART2 neural network is described. At the beginning of the process the algorithm is illustrated step by step by mathematical formulas and it is shown how individual vector changes its values during the training. The network supports clustering by using competitive le...
The ASTM D-5236 method was proven to be the closest to true boiling point (TBP) of heavy oil distillation data. That is why 60 heavy oils have been characterized for their distillation charactersistics by ASTM D-5236 and ASTM D-1160 in the Research laboratory of LUKOIL Neftochim Burgas and an attempt was made to find an accurate method to convert A...
In this research is presented the use of pattern neural network as a useful tool for face recognition. It is described how neural networks can help us in a situation where the main aim is recognizing human faces. The method is described in terms of generalized nets. The training data includes images of the faces. Each of these pictures has been pro...
This paper describes the simulation of generalized net model of ART2 neural network. The article represents the test process of learned network and in order to achieve it the learned bottom-up and top-down weight values are applied in two tokens. When all tokens are initialized the network is ready to be started and the input vector is presented in...
The artificial neural networks (ANN) are a tool that can be used for object recognition and identification. However, there are certain limits when we may use ANN, and the number of the neurons is one of the major parameters during the implementation of the ANN. On the other hand, the bigger number of neurons slows down the learning process. In our...
The generalized net model, presented in the following pages, is implemented in an embedded system, with a purpose to be mounted on a self-moving robot, in order to scan the 2D (or 3D) space around. The system uses a combined 3D accelerometer-magnetometer sensor and an ultrasonic sensor.
Neural networks are a mathematical model for solving problems that uses the structure of the human brain. One of the mostly used kinds of neural networks, the multilayer perceptron (MLP), has been modelled with various tools. In the first model we discussed the MLP, we approached the problem by modelling neural networks in terms of index matrices (...
The paper considers the problem of face recognition from images with application of neural networks. In order to achieve face recognition, different methods and algorithms are combined. At the beginning of the process the correct samples are chosen and several filters are applied on them. After these processes are done the ART2 network is divided i...
The paper presents the generalized net model that describes the processes in the Levenberg–Marquardt algorithm. It is a very useful tool for training of neural network.
The artificial neural networks are inspired by biological properties of human and animal brains. One of the neural networks type is called ART [4]. The abbreviation of ART stands for Adaptive Resonance Theory that has been invented by Stephen Grossberg in 1976 [5]. ART represents a family of Neural Networks. It is a cognitive and neural theory that...
In this paper we present intuitionistic fuzzy neural network as a part of generalized net model of multi-sensorial integrated systems for early detection of forest fires. Many information and data sources have been used, including infrared images, visual images, sensors data, and geographic data bases. One of the main purpose is using of the intell...
Many objects and processes inspired by the nature have been recreated by the scientists. The inspiration to create a Multilayer Neural Network came from human brain as member of the group. It possesses complicated structure and it is difficult to recreate, because of the existence of too many processes that require different solving methods. The ai...
Neural networks are a mathematical model for solving problems, that uses the structure of human brain. One of the mostly used kinds of neural networks, the Multilayer perceptron (MLP), has been modelled with various tools. Here, starting with the MLP, we approach the problem by modelling neural networks in terms of index matrices (IMs). The work in...
Using the approach of the game method for modelling, a model of oil transformation in marine environment is constructed. It is an illustration of the possible applications of the method.
We construct here a Generalized Net (GN) that represents the functioning and the results of the work of real processes and simultaneously – the processes of their control and optimization on the basis of different suitably chosen hierarchical neural networks solving concrete optimization procedures and using information generated in the GN.
The proposed GN model presents the functioning of recurrent neural networks. Here we discuss the Elman network and the 'backpropagation' algorithm for learning. In comparison with other types of neural networks, here we describe the process in its temporal development. In a series of papers, we have described many different neural networks using th...
In the paper we investigate how to apply some data mining techniques for clustering and classification the assessment of the different publications and articles. For this aim we propose to use neural network and decision tree to analyze given collection of data. We use the Intuitionistic fuzzy estimation as an input vector for the self organizing m...
A new procedure for de-i-fuzzification of intuitionistic fuzzy sets is described.
The used generalized net will give us a possibility for parallel optimization of multilayer perceptron based on assigned training pairs with momentum backpropagation algorithm. Here we propose optimization with time limit.
The paper presented development includes hardware which provides the necessary currents and voltages, and application software for processing and graphic displaying the results. Generalized nets are proper tool for the modelling the dynamic structures. Here we design a generalized net that we can use for model this type structures.
It is shown that the game method for modelling can be used for a representation of the neural networks.
Social networks have become a very important part of nowadays life. Many people proposed social network models in the recent years, and most of them focus on classification coefficient. Social networks have been central to some of the most influential theories in sale marketing. For researchers interested in exploring social networks and their rela...
The present paper describes the process of scanning, preprocessing, learning and recognition of the iris. For the purpose we use generalized nets. The model describes an algorithm for the entire process, beginning with scanning, preprocessing of the pictures, and preparing the neural network. In the final stage, the neural network recognizes the di...
The paper presents a neural network that performs clustering and recognition of
petroleum fractions. For the purpose of practical implementation, the degrees of petroleum
probes’ affiliation to the clusters are measured using intuitionistic fuzzy estimations. The entire
process of petroleum identification is modelled using a generalized net.
In this paper a generalized net model has been constructed for preprocessing a digital image with the help of operators over fuzzy, intuitionistic fuzzy and second type intuitionistic fuzzy sets. The enhanced images are compared and the best one among them is chosen for recognition with the data base images. Here, the model is illustrated with the...
The proposed GN model presents the functioning of the Nonlinear Autoregressive Network with Exogenous Inputs O{ARX). This is the one of the recurrent dynamic neural networks with feedback connection with several layers. Instead of other types neural network now we are describing the process during the time.
In this paper we used a generalized net which gives a possibility for parallel optimization of multilayer neural networks. For training the backpropagation algorithm with momentum was considered. We proposed a generalized net model of parallel training of two neural networks with different architectures. The difference between the networks is in th...
The present paper describes the process of the administrative servicing in a digital university. For
the purpose we use Generalized Nets. The opportunity to use GNs as a tool for modelling such
processes is also analyzed. The model utilizes the theory of intuitionistic fuzzy sets.
In a series of research papers, the authors study some of the most important processes of functioning of universities and
construct their Generalized Net (GN) models. The main focus in this paper is to analyse the process of the production of course
timetables in a digital university and to evaluate their quality. The opportunity of using GNs as a...
The Ad Hoc networks are very important for provided LAN services. Ad hoc use wireless communication and very often have problem for creation and support of the transfer of the information. A purpose of the present work is the development of the neural network giving intuitionistic fuzzy estimation for reliability of the connection.
The present paper describes the process of the European Awareness Scenario Workshop Method. For the purpose we use Generalized Nets. The opportunity of using GNs as a tool for modelling such process is analyzed as well. The model uses intuitionistic fuzzy sets for determine estimation for a creation of the masterplan.
The paper presents a neural network that evaluates the students' answers based on defined set criterions. To be involved in practice there are used evaluation in intuitionistic fuzzy form about the students' knowledge. This is proper to be used as a basic element for e-learning systems' building. Introduction Within the context of e-learning, the i...
This article develops an intuitionistic fuzzy estimation of the availability of a radio signal from a mobile unit to the base station in a shadowing and multipath propagation environment. This estimation method provides a way to optimally manage the switching strategy of the base station depending on the signal level of the different mobile units s...
In previous paper we introduced feed forward neural network changed into the definition of the Intuitionistic Fuzzy Logic (IFL). All its parameters used IFL characteristics. In this paper a generalized net model us constructed that introduces work in the IFFFNN.
In a series of research papers, the authors have studied some of the most important models of contemporary universities, such as: the research university, the entrepreneurial university and the digital university and construct their generalized net (GN) models. This paper is based on the case-studies of Sofia University, the Technical University of...
In this paper we introduce GN-model of the work of the electronic payment via Internet. The model can be used for the optimization and following the networkpsilas behaviour in future.
The present paper describes the process of the verification by Iris scanning. For the purpose we use Generalized Nets. The model describes an algorithm using phase-based image-an image matching technique using only the phase components.
In a series of research papers, the authors have studied some of the most important models of a contemporary universities, such as: research university, entrepreneurial university and digital university and construct their Generalized Net (GN) models. The paper is based on the case-studies of Sofia University, Technical University of Munich and Uni...
The present paper describes the process of ordering of university subjects. The evaluations corresponding to the expert's reliability score and all expert opinions about the evaluation of the relation between different disciplines utilize the theory of intuitionistic fuzzy sets.
Ad-hoc networks are created by definition on demand without any infrastructure. They are often considered as an extension of the range of Internet access points, providing multihop wireless access to them. This paper tries to examine the situation where there are exist several Internet access points in a single ad-hoc network. We present a generali...
The used generalized net will give us a possibility for parallel optimization of feed-forward neural network based on assigned training pairs and time limits.
We introduce a GN-model with intuitionistic fuzzy estimations of transmit routines of the Bluetooth communication in part data traffic.
In this paper we introduce e GN-model of the work of Learning Vector Quantization neural networks. The model can be used for the optimization and following the network's behavior in future.
The neural networks self-organizing maps have got a multitude of applications. The introduced generalized net gives a possibility for optimization of the neural network allowing determination of its optimum architecture based on time limits for its learning.
In this article we introduce an accelerated modeling method for neural network education by a back-propagation algorithm, changing the description of the standard algorithm with the aim of accelerated education of the neural network. Generalized net models of an accelerating backpropagation algorithm are constructed.
The article presents of accelerating neural network learning by the Back Propagation algorithm and one of its fastest modifications – the Levenberg–Marqurdt method. The learning is accelerated by introducing the ‘single-direction’ coefficient of the change of x for calculating its new values (the number of iterations is decreased by approximately 3...
Some metals are absolutely necessary for the welfare of the human’s vital activity and relate to so called biogene elements. Other metals – so called toxic metals, provoke opposite effects – getting into a living organism they cause poisoning and death. About their examination however are necessary many experiments and compound mathematical methods...
Nullor models of active devices are widely used for simulation of linear active networks. The fault arising of a passive network element causes definite changes of the value parameters (voltages and currents) of active elements - independent voltage and current sources and norators. These changes can be used as indicators for the fault diagnosis of...
In the previous article [the authors, “Modeling the work of self organising neural networks with generalized networks”, in: K. Atanassov et al. (eds.), Third international workshop on intuitionistic fuzzy sets and generalized nets, Warsaw, Poland, September 7–8, 2003. Issues in Intuitionistic Fuzzy Sets and Generalized Nets 3, 57–63 (2006; Zbl 1162...
Summary form only given. Stream measuring of elemental parts at high energy operations has always created a great interest on behalf of the direct custumers and researchers. Elementary parts dispersion from accelerators, has created various questions, concerning the input energy, its control, and the final outcome in terms of outcome parameters. Th...
The used generalized net will give us a possibility for parallel optimization of a feed-forward neural network based on assigned
training pairs with variable learning rate backpropagation algorithm with time limit.
The present paper describes the process of serving transport flow in the terminals. The model can be used for process optimization.
In this paper we introduce a GN-model of the Short Message Service. The model can be used for the optimization and following the network's behavior in future.
The present paper describes the process work of the car's electronic systems. For the purpose we use Generalized Nets. The model represents few different electronic systems in car. Introduction Anti-Blocker System is a safety system on motor vehicles which prevents the wheels from locking while braking. A non-locking braking system allows the drive...
A generalized net is used to construct a model which describes the process of changing of the location of the network subsystem.
