Contents of Volume 24 (2014)

1/2014 2/2014 3/2014 4/2014 5/2014


  • [1] Pelikán E. (CZ)
    Tutorial:Forecasting of processes in complex systems for real-world problems, 567-589

    First page   Full text     DOI: 10.14311/NNW.2014.24.032

    Abstract: This tutorial is based on modification of the professor nomination lecture presented two years ago in front of the Scientific Council of the Czech Technical University in Prague [16]. It is devoted to the techniques for the models developing suitable for processes forecasting in complex systems. Because of the high sensitivity of the processes to the initial conditions and, consequently, due to our limited possibilities to forecast the processes for the long-term horizon, the attention is focused on the techniques leading to practical applications of the short term prediction models. The aim of this tutorial paper is to bring attention to possible difficulties which designers of the predicting models and their users meet and which have to be solved during the prediction model developing, validation, testing, and applications. The presented overview is not complete, it only reflects the author’s experience with developing of the prediction models for practical tasks solving in banking, meteorology, air pollution and energy sector. The paper is completed by an example of the global solar radiation prediction which forms an important input for the electrical energy production forecast from renewable sources. The global solar radiation forecasting is based on numerical weather prediction models. The time-lagged ensemble technique for uncertainty quantification is demonstrated on a simple example.

  • [2] Sung Ho Jang, Hyeok Gyu Kwon (Korea)
    Neural connectivity of the amygdala in the human brain: A diffusion tensor imaging study, 591-600

    First page   Full text     DOI: 10.14311/NNW.2014.24.033

    Abstract: Several diffusion tensor imaging (DTI) studies have reported on the anatomical neural tracts between the amygdala and specific brain regions. However, no study on the neural connectivity of the amygdala has been reported. In the current study, using probabilistic DTI tractography, we attempted to investigate the neural connectivity of the amygdala in normal subjects. Forty eight healthy subjects were recruited for this study. A seed region of interest was drawn at the amygdala using the FMRIB Software Library based on probabilistic DTI tractography. Connectivity was defined as the incidence of connection between the amygdala and each brain region at the threshold of 1 and 5 streamlines. The amygdala showed 100% connectivity to the hippocampus, thalamus, hypothalamus, and medial temporal cortex regardless of the thresholds. In contrast, regarding the thresholds of 1 and 5 streamlines, the amygdala showed high conncetivity (over 60%) to the globus pallidus (100% and 92.7%), brainstem (83.3% and 78.1%), putamen (72.9% and 63.5%), occipito-temporal cortex (72.9% and 67.7%), orbitofrontal cortex (70.8 and 43.8%), caudate nucleus (63.5% and 45.8%), and ventromedial prefrontal cortex (63.5% and 31.3%), respectively. The amygdala showed high connectivity to the hippocampus, thalamus, hypothalamus, medial temporal cortex, basal ganglia, brainstem, occipito-temporal cortex, orbitofrontal cortex, and ventromedial prefrontal cortex. We believe that the methods and results of this study provide useful information to clinicians and researchers studying the amygdala.

  • [3] Lazzús J. A., Salfate I., Montecinos S. (Chile)
    Hybrid neural network–particle swarm algorithm to describe chaotic time series, 601-617

    First page   Full text     DOI: 10.14311/NNW.2014.24.034

    Abstract: An artificial neural network (ANN) based on particle swarm optimization (PSO) was developed for the time series prediction. This hybrid ANN+PSO algorithm was applied on Mackey-Glass series in the short-term prediction x(t+6) and the long-term prediction x(t+84), from the current value x(t) and the past values: x(t-6), x(t-12), x(t-18). Four cases were studied, alternating the timedelay parameter as 17 or 30. Also, the first four largest Lyapunov exponents were obtained for different time-delay. Simulation shows that this ANN+PSO method is a very powerful tool for making prediction of chaotic time series.

  • [4] Acir N., Mengüc E. C.(Turkey)
    Lyapunov theory based adaptive learning algorithm for multilayer neural networks, 619-636

    First page   Full text     DOI: 10.14311/NNW.2014.24.035

    Abstract: This paper presents a novel weight updating algorithm for training of multilayer neural network (MLNN). The MLNN system is first linearized and then the design procedure is proposed as an inequality constraint optimization problem. A well selected Lyapunov function is suitably determined and integrated into the constraint function for satisfying asymptotic stability in the sense of Lyapunov. Thus, the convergence capability of training algorithm is improved by using a new analytical adaptation gain rate which has the ability to adaptively adjust itself depending on a sequential square error rate. The proposed algorithm is compared with two types of backpropagation algorithms and a Lyapunov theory based MLNN algorithm on three benchmark problems which are XOR, 3-bit parity, and 8-3 encoder. The results are compared in terms of number of learning iterations and computational time required for a specified convergence rate. The results clearly indicate that the proposed algorithm is much faster in convergence than other three algorithms. The proposed algorithm is also comparatively tested on a real iris image database for multiple-input and multiple-output classification problem and the effect of adaptation gain rate for faster convergence and higher performance is verified.

  • [5] Tayfur G., Bektas B., Duvarci Y. (Turkey)
    Significance of rent attributes in prediction of earthquake damage in Adapazari, Turkey, 637-653

    First page   Full text     DOI: 10.14311/NNW.2014.24.036

    Abstract: This paper analyses rent-based determinants of earthquake damage from an urban planning perspective with the data gathered from Adapazari, Turkey, after the disaster in 1999 Eastern Marmara Earthquake (EME). The study employs linear regression, log-linear regression, and artificial neural networks (ANN) methods for cross-verification of results and for finding out the significant urban rent attribute(s) responsible for the damage. All models used are equally capable of predicting the earthquake damage and converge to similar results even if the data are limited. Of the rent variables, the physical density is proved to be especially significant in predicting earthquake damage, while the land value contributes to building resistance. Thus, urban rent can be the primary tool for planners to help reduce the fatalities in preventive planning studies.

  • [6] Aktepe A., Ersöz S., Lüy M. (Turkey)
    Welding process optimization with artificial neural network applications, 655-670

    First page   Full text     DOI: 10.14311/NNW.2014.24.037

    Abstract: Correct detection of input and output parameters of a welding process is significant for successful development of an automated welding operation. In welding process literature, we observe that output parameters are predicted according to given input parameters. As a new approach to previous efforts, this paper presents a new modeling approach on prediction and classification of welding parameters. 3 different models are developed on a critical welding process based on Artificial Neural Networks (ANNs) which are (i) Output parameter prediction, (ii) Input parameter prediction (reverse application of output prediction model) and (iii) Classification of products. In this study, firstly we use Pareto Analysis for determining uncontrollable input parameters of the welding process based on expert views. With the help of these analysis, 9 uncontrollable parameters are determined among 22 potential parameters. Then, the welding process of ammunition is modeled as a multi-input multi-output process with 9 input and 3 output parameters. 1st model predicts the values of output parameters according to given input values. 2nd model predicts the values of correct input parameter combination for a defect-free weld operation and 3rd model is used to classify the products whether defected or defect-free. 3rd model is also used for validation of results obtained by 1st and 2nd models. A high level of performance is attained by all the methods tested in this study. In addition, the product is a strategic ammunition in the armed forces inventory which is manufactured in a limited number of countries in the world. Before application of this study, the welding process of the product could not be carried out in a systematic way. The process was conducted by trialand- error approach by changing input parameter values at each operation. This caused a lot of costs. With the help of this study, best parameter combination is found, tested, validated with ANNs and operation costs are minimized by 30%.


  • [1] Křížek M.,Somer L. (CZ, USA)
    Tutorial:A critique of the standard cosmological model, 435-462

    First page   Full text     DOI: 10.14311/NNW.2014.24.026

    Abstract: According to the standard cosmological model, 27 % of the Universe consists of some mysterious dark matter, 68 % consists of even more mysterious dark energy, whereas only less than 5 % corresponds to baryonic matter composed from known elementary particles. The main purpose of this paper is to show that the proposed ratio 27 : 5 between the amount of dark matter and baryonic matter is considerably overestimated. Dark matter and partly also dark energy might result from inordinate extrapolations, since reality is identified with its mathematical model. Especially, we should not apply results that were verified on the scale of the Solar System during several hundreds of years to the whole Universe and extremely long time intervals without any bound of the modeling error.

  • [2] Mizera P., Pollak P. (CZ)
    Robust neural network-based estimation of articulatory features for Czech, 463-478

    First page   Full text     DOI: 10.14311/NNW.2014.24.027

    Abstract: The article describes a neural network-based articulatory feature (AF) estimation for the Czech speech. First, the relationship between AFs and a Czech phone inventory is defined, and then the estimation based on the MLP neural networks is done. The usage of several speech representations on the input of the MLP classifiers is proposed with the purpose to obtain a robust AF estimation. The realized experiments have proved that an ANN- based AF estimation works very reliably especially in a low noise environment. Moreover, in case the number of neurons in a hidden layer is increased and if the temporal context DCT-TRAP features are used on the input of the MLP network, the AF classification works accurately also for the signals collected in the environments with a high background noise.

  • [3] Ching-Hui Shih, Sin-Jin Lin, Ming-Fu Hsu (Taiwan)
    Detection of financial information manipulation by an ensemble-based mechanism, 479-500

    First page   Full text     DOI: 10.14311/NNW.2014.24.028

    Abstract: Complicated financial information manipulation, involving heightened offender knowledge of transactional procedures, can be damaging to the reputations of corporations and the auditors, as well as cause serious turbulence in financial markets. Unfortunately, most incidents of financial information manipulation involve higher level managers who are truly knowledgeable and comprehend the limitations of standard auditing procedures. Thus, there is an urgent need for additional detection mechanisms to prevent financial information manipulation. To address this problem, the author proposes an ensemble-based mechanism (EM) consisting of feature selection and extraction ensemble and extreme learning machine (ELM). The model not only counters the redundancy-removing problem, but also gives direction to auditors who need to allocate limited audit resources to abnormal client relationships during the auditing procedure and protect the CPA firms’ reputation. The experimental results demonstrate that the model is a promising alternative for detecting financial information manipulation, and one that can ensure both the confidence of investors and the stability of financial markets.

  • [4] Almasi O. N., Akhtarshenas E., Rouhani M. (Iran)
    An efficient model selection for SVM in real–world datasets using BGA and RGA, 501-520

    First page   Full text     DOI: 10.14311/NNW.2014.24.029

    Abstract: Support vector machine (SVM) has become one of the most popular machine-learning methods during the last years. The design of an efficient model and the proper adjustment of the SVMs parameters are integral to reducing the testing time and enhancing performance. In this paper, a new bipartite objective function consisted of the sparseness property and generalization performance is proposed. Since the proposed objective function is based on selecting fewer numbers of the support vectors, the model complexity is reduced while the performance accuracy remains at an acceptable level. Due to the model complexity reduction, the testing time is decreased and the ability of SVM in practical applications is increased Moreover, to prove the performance of the proposed objective function, a comparative study was carried out on the proposed objective function and the conventional objective function, which is only based on the generalization performance, using the Binary Genetic Algorithm (BGA) and Real-valued vectors GA (RGA). The effectiveness of the proposed cost function is demonstrated based on the results of the comparative study on four real-world datasets of UCI database.

  • [5] Naseem M. T., Qureshi I. M., Atta-ur-Rahman, Muzaffar M. Z. (Pakistan)
    Image selection criteria for embedding desired capacity using FRBS, 521-538

    First page   Full text     DOI: 10.14311/NNW.2014.24.030

    Abstract: Data hiding methods are used to carry information from one place to another. Digital watermarking is one of the data hiding methods. Imperceptibility and capacity are the conflicting parameters in digital watermarking. The more the embedded information, lower the imperceptibility and vice versa. Imperceptibility factor (IF) is measured as peak signal to noise ratio (PSNR) of the image after embedding information. No such schemes exist in the literature in which an image can be chosen that may carry a desired capacity, while keeping imperceptibility as high as possible. In this scheme a two stage fuzzy rule based system (FRBS) is designed to choose the image among the list that is capable of holding desired capacity while achieving high imperceptibility at the same time. Validity of the proposed scheme is checked through simulation results of different types of images like natural and medical. Moreover, the proposed scheme is also robust against JPEG compression attack.

  • [6] Shin-Ying Huang, Rua-Huan Tsaih, Wan-Ying Lin (Taiwan)
    Feature extraction of fraudulent financial reporting through unsupervised neural networks, 539-560

    First page   Full text     DOI: 10.14311/NNW.2014.24.031

    Abstract: The objective of this study is to apply an unsupervised neural network tool to analyze fraudulent financial reporting (FFR) by extracting distinguishing features from samples of groups of companies and converting them into useful information for FFR detection. This methodology can be used as a decision support tool to help build an FFR identification model or other financial fraud or financial distress scenarios. The three stages of the proposed quantitative analysis approach are as follows: the data-preprocessing stage; the clustering stage, which uses an unsupervised neural network tool known as a growing hierarchical self-organizing map (GHSOM) to cluster sample observations into subgroups with hierarchical relationships; and the feature-extraction stage, which uncovers common features of each subgroup via principle component analysis. This study uses the hierarchal topology mapping ability of a GHSOM to cluster financial data, and it adopts principal component analysis to determine common embedded features and fraud patterns. The results show that the proposed three-stage approach is helpful in revealing embedded features and fraud patterns, using a set of significant explanatory financial indicators and the proportion of fraud. The revealed features can be used to distinguish distinctive groups.


  • [1] Michail Masikos, Konstantinos Demestichas, Evgenia Adamopoulou, Michael Theologou (Greece)
    Reliable vehicular consumption prediction based on machine learning, 333-342

    First page   Full text     DOI: 10.14311/NNW.2014.24.019

    Abstract: A robust prediction model is developed for reliably estimating vehicular consumption. This model is distinguished from other models proposed so far for the following reasons: it detects the factors contributing into vehicular consumption, it applies machine learning functionality for approximating the nonlinearities and the specificities between the contributing factors, and it is capable of implicitly adapting to the characteristics of the vehicle, the road network and the contextual conditions through its learning process. The authors validated its efficiency by applying it on measurements collected during a data acquisition campaign, which was performed by a fully electric vehicle (FEV) in an urban road network.

  • [2] Mehmet Serhat Odabas, Navaratnam Leelaruban, Halis Simsek, G. Padmanabhan (Turkey, USA)
    Quantifying impact of droughts on barley yield in North Dakota, USA using Multiple Linear Regression and Artificial Neural Network, 343-356

    First page   Full text     DOI: 10.14311/NNW.2014.24.020

    Abstract: This research investigated the effect of different drought conditions on Barley (Hordeum vulgare L.) yield in North Dakota, USA, using Multiple Linear Regression (MLR) and Artificial Neural Network (ANN) methods. Though MLR method is widely used, the ANN method has not been used in the past to investigate the effect of droughts on barley yields to the best of authors’ knowledge. It is found from this study that the ANN model performs better than MLR in estimating barley yield. In this paper, the ANN is proposed as a viable alternative method or in combination with MLR to investigate the impact of droughts on crop yields.

  • [3] Guijun Wang, Xiaoping Li (China)
    Construction of the polygonal fuzzy neural network and its approximation based on K-integral norm, 357-376

    First page   Full text     DOI: 10.14311/NNW.2014.24.021

    Abstract:The concept of an n-equidistant polygonal fuzzy number is introduced to avoid the complexity of the operations between fuzzy numbers. Firstly, the properties of linear operations and the convergence of n-equidistant polygonal fuzzy numbers are discussed, the method how to change a fuzzy number into an n-equidistant polygonal fuzzy number is shown. Next, for given a µ-integrable polygonal fuzzy valued function, an n-equidistant polygonal fuzzy valued function is constructed. By introducing the definition of K-quasi-additive integral and Kintegral norm, the universal approximation of polygonal fuzzy neural network are studied. The final result indicates that the polygonal fuzzy neural network still possess universal approximation to an integrable system.

  • [4] Yunquan Ke, Chunfang Miao (China)
    Exponential stability of periodic solutions for inertial Cohen-Grossberg-type neural networks, 377-394

    First page   Full text     DOI: 10.14311/NNW.2014.24.022

    Abstract:In this paper, the exponential stability of periodic solutions for inertial Cohen-Grossberg-type neural networks are investigated. First, by properly chosen variable substitution the system is transformed to first order differential equation. Second, some sufficient conditions which can ensure the existence and exponential stability of periodic solutions for the system are obtained by using constructing suitable Lyapunov function and differential mean value theorem, applying the analysis method and inequality technique. Finally, two examples are given to illustrate the effectiveness of the results.

  • [5] Yen-Bin Chen, Yung-Lung Lee, Shou-Jen Hsu, Yi-wei Chen (Taiwan, R.O.C.)
    Design of active heat dissipation system for adaptive wavelet neural network control, 395-410

    First page   Full text     DOI: 10.14311/NNW.2014.24.023

    Abstract:This paper develops an Adaptive Wavelet Neural Network Control (AWNNC) algorithm for radar active heat dissipation system. The radar core processor belongs to a highly precision component which consists of the electronic device of radio frequency integrated circuit (RFIC) with high power and high performance. The radar core processor should be operated in a narrowly closed environment without convection, which will increase the heat sink effect inside the core processor and further affect its reliability and life-time. The AWNNC comprises a wavelet neural network (WNN) controller and a robust compensator. The WNN controller is a principal tracking controller which is utilized to mimic an ideal controller; and the parameters of WNN are online tuned by the derived adaptation laws based on the gradient descent method. The robust compensator is designed to dispel the approximation error between the ideal controller and the WNN controller, thus the asymptotic stability of the closed–loop system can be achieved. Based on National Instruments-PCI extensions for Instrumentation (NI-PXI) system, combined the Thermo Electric Cooler (TEC) with a duct heater, active heat dissipation intelligent control system is designed to fix the problem of heat dissipation in long distance in a narrowly closed environment without convection. According to the amount of thermal source and thermal energy, the smart control system can help to adjust the rate of heat dissipation by taking advantage of an adaptive control so that the performance of heat dissipation may be accumulated by its numbers. Last but not least, compared the traditional analog circuit controller with adaptive wavelet neural network controller, the research proves that its proposed active heat dissipation intelligent control system can reach an excellent and accurate temperature control. Speaking more precisely, adaptive wavelet neural network controller can be easily adaptive to any environment. It is equipped with a good capability of tracking and searching; and in terms of the effect of temperature control, it never actually jitters due to an input of voltage saturation compared with traditional analog circuit controller. All these can make chips able to adjust its adaptive rate of heat dissipation in accordance with the thermal source of the chips in a narrowly closed environment without convection.

  • [6] Svítek M. (CZ)
    Iterated non-linear regression, 411-420

    First page   Full text     DOI: 10.14311/NNW.2014.24.024

    Abstract: The paper presents iterated algorithm for parameter estimation of non-linear regression model. The non-linear model is firstly approximated by a polynomial. Afterwards, parameter estimation based on measured data is taken as the initial value for the proposed iterated algorithm. As the estimation method, the well-known Least Square Estimation (LSE), artificial neural networks (ANN) or Bayesian methodology (BM) can be used. With respect to the knowledge of initial parameters the measured data are transformed to meet best the non-linear regression criteria (orthogonal data projection). The original and transformed data are used in the next step of the designed iterated algorithm to receive better parameter estimation. The iteration is repeated until the algorithm converges into a final result. The proposed methodology can be applied on all non-linear models that could be approximated by a polynomial function. The illustrative examples show the convergence of the designed iterated algorithm.

  • [7] Ding Gang, Lei Da, Zhong Shisheng (China)
    Time series prediction using convolution sum discrete process neural network, 421-432

    First page   Full text     DOI: 10.14311/NNW.2014.24.025

    Abstract: A convolution sum discrete process neural network (CSDPNN) is proposed. CSDPNN utilizes discrete samples as inputs directly and employs convolution sum to simulate the process inputs so as to deal with the time accumulation existing in many time series. Without the procedures of fitting the discrete samples into continuous functions to generate inputs and then to expand the input functions by basis functions, CSDPNN is better understandable and is with less precision reduction compared with process neural network (PNN) with function inputs. The approximation capacity of CSDPNN is analyzed in this paper, and it proved that CSDPNN can approximate PNN and has approximation capacity not worse than traditional artificial neural network (ANN). Finally, CSDPNN, PNN and ANN are utilized to predict the Logistic chaos time series and the iron concentration in the lubrication oil of aircraft engine, and the application test results indicate that CSDPNN performs better than PNN and ANN given the same conditions.


  • [1] Vesely A. (CZ)
    Information content of association rules, 231-248

    First page   Full text     DOI: 10.14311/NNW.2014.24.014

    Abstract: Database records can be often interpreted as state descriptions of some world, system or generic object, states of which occur independently and are described by binary properties. If records do not contain missing values, then there exists close relationship between association rules and propositions about state properties. In data mining we usually get a lot of association rules with large confidence and large support. Since their interpretation is often cumbersome, some quantitative measure of their “informativeness” would be very helpful. The main aim of the paper is to define a measure of the amount of information contained in an association rule. For this purpose we make use of the tight correspondence between association rules and logical implications. At first a quantitative measure of information content of logical formulas is introduced and studied. Information content of an association rule is then defined as information content of the corresponding logical implication in the situation when no knowledge about dependence among properties of world states is at our disposal. The intuitive meaning of the defined measure is that the association rule that allows more appropriate correction of the distribution of world states, acquired under unfair assumption of independence of state properties, contains also larger amount of information. The more appropriate correction here means a correction of the current probability distribution of states that leads to the distribution that is closer to the true distribution in the sense of Kullback-Leibler divergence measure.

  • [2] Kumar V., Venkatesh K., Tiwari R. P. (India)
    A neurofuzzy technique to predict seismic liquefaction potential of soils, 249-266

    First page   Full text     DOI: 10.14311/NNW.2014.24.015

    Abstract: Liquefaction potential is a scientific assessment parameter to assess liquefaction of medium to fine grained cohesion-less soil due to earthquake shaking. In this paper alternative liquefaction potential prediction models have been developed using adaptive neuro fuzzy inference system (ANFIS) and multiple linear regression (MLR) technique. Geological survey of the study area was performed and forty locations were identified to perform standard penetration test (SPT). Disturbed and undisturbed soil samples were collected from the borehole to execute the laboratory tests. The bore-log datasets were used for determining liquefaction potential of the cohesion-less soils. The analytical approach by Idriss and Boulanger (I & B) has been applied initially to estimate liquefaction potential of soil on the basis of standard penetration test datasets obtained from the field investigations. To develop the ANFIS models 101 datasets were collected and incorporated for the development of fuzzy neural network models. Multiple linear regression (MLR) models have also been developed and the results were compared with neuro-fuzzy models. Based on obtained results it can be stated that the developed adaptive neuro fuzzy inference system models have better prediction ability to predict liquefaction potential with satisfactory level of confidence and can be used as an alternative tool.

  • [3] Aslantas G., Gürgen F., Salah A. A. (Turkey)
    GA-NN approach for ECG feature selection in rule based arrhythmia classification, 267-283

    First page   Full text     DOI: 10.14311/NNW.2014.24.016

    Abstract: Computer-aided ECG analysis is very important for early diagnosis of heart diseases. Automated ECG analysis integrated with experts' opinions may provide more accurate and reliable results for detection of arrhythmia. In this study, a novel genetic algorithm-neural network (GA-NN) approach is proposed as a classifier, and compared with other classification methods. The GA-NN approach was shown to perform better than alternative approaches (e.g. k-nn, SVM, naive Bayes, Bayesian networks) on the UCI Arrythmia and the novel TEPAS ECG datasets, where the GA resulted in a feature reduction of 95%. Based on the selected features, several rule extraction algorithms are applied to allow the interpretation of the classification results by the experts. In this application, the accuracy and interpretability of results are more important than processing speed. The results show that neural network based approaches benefit greatly from dimensionality reduction, and by employing GA, we can train the NN reliably.

  • [4] Jinpeng Chen, Yu Liu, Zhenyu Wu, Ming Zou, Deyi Li (China)
    Recommending interesting landmarks in photo sharing sites, 285-308

    First page   Full text     DOI: 10.14311/NNW.2014.24.017

    Abstract: With the rapid development of location-acquisition technologies (GPS, GSM networks, etc.), more and more unstructured, geo-referenced data are accumulated on the Web. Such abundant location-based data imply, to some extent, users’ interests in places, so these data can be exploited for various location-based services, such as tour recommendation. In this paper, we demonstrate that, through utilizing the location data from a popular photo sharing web site such as Flickr, we can explore interesting landmarks for recommendations. We aim to generate personalized landmark recommendations based on geo-tagged photos for each user. Meanwhile, we also try to answer such a question that when we want to go sightseeing in a large city like Beijing, where should we go? To achieve our goal, first, we present a data field clustering method (DFCM), which is a density-based clustering method initially developed to cluster point objects. By using DFCM, we can cluster a large-scale geo-tagged web photo collection into groups (or landmarks) by location. And then, we provide more friendly and comprehensive overviews for each landmark. Subsequently, we present an improved user similarity method, which not only uses the overview semantic similarity, but also considers the trajectory similarity and the landmark trajectory similarity. Finally, we propose a personalized landmark recommendation algorithm based on the improved user similarity method, and adopt a TF-IDF like strategy to produce the nontrivial landmark recommendation. Experimental results show that our proposed approach can obtain a better performance than several state-of-the-art methods.

  • [5] Guofang Nan, Zhongnan Chen, Minqiang Li, Liang Huang, Ajith Abraham (China)
    Distributed deployment algorithm based on boundary expansion and virtual force for mobile sensor networks, 309-332

    First page   Full text     DOI: 10.14311/NNW.2014.24.018

    Abstract: Optimization of sensors’ position is a challenging problem in wireless sensor networks since the processing process significantly affects energy consumption, surveillance ability and network lifetime. Vectorbased algorithm (VEC) and Voronoi-based algorithm (VOR) are two existing approaches. However, VEC is sensitive to initial deployment, while VOR always moves to the coverage holes. Moreover, the nodes in a network may oscillate for a long time before they reach the equilibrium state. This paper presents an initially central deployment model that is cost effective and easy to implement. In this model, we present a new distributed deployment algorithm based on boundary expansion and virtual force (BEVF). The proposed scheme enables nodes to move to the boundary rapidly and ultimately reach equilibrium quickly. For a node, only the location of its nearby nodes and boundary information are needed in the algorithm, thereby avoiding communication cost for transmitting global information. The distance threshold is adopted to limit node movement and to avoid node oscillations. Finally, we compare BEVF with existing algorithms Results show that the proposed algorithm achieves a much larger coverage and consumes lower energy.


  • [1] Bessou S., Touahria M. (Algeria)
    An accuracy-enhanced stemming algorithm for Arabic information retrieval, 117-128

    First page   Full text     DOI: 10.14311/NNW.2014.24.007

    Abstract: This paper provides a method for indexing and retrieving Arabic texts, based on natural language processing. Our approach exploits the notion of template in word stemming and replaces the words by their stems. This technique has proven to be effective since it has returned significant relevant retrieval results by decreasing silence during the retrieval phase. Series of experiments have been conducted to test the performance of the proposed algorithm ESAIR (Enhanced Stemmer for Arabic Information Retrieval). The results obtained indicate that the algorithm extracts the exact root with an accuracy rate up to 96% and hence, improving information retrieval.

  • [2] Ramesh Kumar T., Rajendran I. (India)
    Mass loss prediction of newly developed aluminium-based alloys using artificial neural network, 129-142

    First page   Full text     DOI: 10.14311/NNW.2014.24.008

    Abstract: The purpose of this study is to predict the mass loss of newly developed aluminium based alloy. Two different alloys are prepared by cladding process and the sliding friction and wear properties of this alloy against high carbon high chromium steel are investigated at different normal loads (50 N, 60 N and 70 N) under different sliding distances. Tests are carried at a constant speed of 1 m/sec under oil lubricated conditions by preheating the circulating engine oil 20w40 at a temperature of 800C. The mass losses are measured and recorded for every interval. An artificial neural network (ANN) model is developed to predict the mass loss of newly developed aluminium-based alloy. It is observed that the predicted values have shown good agreement with experimental values with a correlation coefficient of 0.999973. This model can also be used to predict the mass loss of any material.

  • [3] Zjavka L. (CZ)
    "Aladin" weather model local revisions using the differential polynomial neural network, 143-156

    First page   Full text     DOI: 10.14311/NNW.2014.24.009

    Abstract: The 48-hour "Aladin" forecast model can predict significant meteorological quantities in a middle scale area. Neural networks could try to replace some statistical techniques designed to adapt a global meteorological numerical forecast model for local conditions, described with real data surface observations. They succeed commonly a cut above problem solutions with a predefined testing data set, which provides bearing inputs for a trained model. Time-series predictions of the very complex and dynamic weather system are sophisticated and not any time faithful using simple neural network models entered only some few variables of their own next-time step estimations. Predicted values of a global meteorological forecast might instead enter a neural network locally trained model, for refine it. Differential polynomial neural network is a new neural network type developed by the author; it constructs and substitutes for an unknown general sum partial differential equation of a system description, with a total sum of fractional polynomial derivative terms. This type of non-linear regression is based on trained generalized data relations, decomposed into many partial derivative specifications. The characteristics of composite differential equation solutions of this indirect type of a function description can facilitate a much greater variety of model forms than is allowed using standard soft-computing methods. This adjective derivative model type is supposed to be able to solve much more complex problems than is usual using standard neural network techniques.

  • [4] Kovanda J., Hozman J., Bradáč J. (CZ)
    Neural tissue response to impact - numerical study of wave propagation at level of neural cells, 157-176

    First page   Full text     DOI: 10.14311/NNW.2014.24.010

    Abstract: In this article, we deal with a numerical solution of the issue concerning one-dimensional longitudinal mechanical wave propagation in linear elastic neural weakly heterogeneous media. The crucial idea is based on the discretization of the wave equation with the aid of a combination of the discontinuous Galerkin method for the space semi-discretization and the Crank-Nicolson scheme for the time discretization. The linearity of the second-order hyperbolic problem leads to a solution of a sequence of linear algebraic systems at each time level. The numerical experiments performed for the single traveling wave and Gauss initial impact demonstrate the high-resolution properties of the presented numerical scheme. Moreover, a well-known linear stress-strain relationship enables us to analyze a high-frequency regime for the initial excitation impact with respect to strain-frequency dependency.

  • [5] Apdullah Yayik, Yakup Kutlu (Turkey)
    Neural Network Based Cryptography, 177-192

    First page   Full text     DOI: 10.14311/NNW.2014.24.011

    Abstract: In this paper, neural network based cryptology is performed. The system consists of two stages. In the first stage, neural network-based pseudo-random numbers (NPRNGs) are generated and the results are tested for randomness using National Institute of Standard Technology (NIST) randomness tests. In the second stage, a neural network-based cryptosystem is designed using NPRNGs. In this cryptosystem, data, which is encrypted by non-linear techniques, is subject to decryption attempts by means of two identical artificial neural networks (ANNs). With the first neural network, non-linear encryption is modeled using relationbuilding functionality. The encrypted data is decrypted with the second neural network using decision-making functionality.

  • [6] Andreas F. V. Roy, Min-Yuan Chengy, Yu-Wei Wu (Indonesia, Taiwan)
    Time dependent evolutionary fuzzy support vector machine inference model for predicting diaphragm wall deflection, 193-210

    First page   Full text     DOI: 10.14311/NNW.2014.24.012

    Abstract: Brace diaphragm walls are commonly used in underground structures in metropolitan areas, where avoiding costly damage to adjacent infrastructure / buildings is critical to project success. It is necessary to make accurate diaphragm wall deflection predictions to ensure actual deflection falls within allowable limits, and thus ensure the safety of both the project and adjacent structures. Numerous studies and approaches, such as empirical, semi-empirical as well as numerical approaches, have addressed excavation-induced deflection in diaphragm walls. Artificial intelligence (AI) has been used recently by several researchers to improve diaphragm wall deflection prediction capabilities. This paper proposes a hybrid artificial intelligence system, namely the evolutionary fuzzy support vector machine inference model for time series data (EFSIMT ), to predict diaphragm wall deflection in deep excavation through the application of historical project data. Simulations were performed on 1,083 instances, segregated into a total of 988 training data sets and 95 test data sets. Validation results show that the EFSIMT achieves higher performance in comparison with Artificial Neural Networks and the Evolutionary Support Vector Machine Inference Model (ESIM). Therefore, EFSIMT has great potential as a predictive tool for diaphragm wall deflection problems and assisting project managers/engineers to ensure safety during the construction process.

  • [7] Asgari H., Kavian Y. S. (Iran)
    Hardware description of digital Hopfield neural networks for solving shortest path problem, 211-230

    First page   Full text     DOI: 10.14311/NNW.2014.24.013

    Abstract: The shortest path problem is an important issue in communication networks which is used by many practical routing protocols. The aim of this paper is to present an intelligent model based on Hopfield neural networks (HNNs) for solving shortest path problem and implement that on Field Programmable Gate Arrays (FPGAs) chips. The Cyclone Π-EP2C70F896C6 FPGA chip from ALTERA Inc. is considered for hardware implementing and VHDL language is employed for hardware description. The synthesizing results show the proposed architecture of neuron is more efficient than relevant neuron model for chip area utilization and consequently improving the maximum operating frequency and power consumption. The proposed router core is employed to find shortest paths in ring, star and mesh communication networks and the results demonstrate the efficiency and superiority of proposed core.


  • [1] L. J. M. Rothkrantz (NL)
    TUTORIAL - Surveillance Angels, 3-29

    First page   Full text     DOI: 10.14311/NNW.2014.24.001

    Abstract: The use of sensor networks has been proposed for military surveillance and environmental monitoring applications. Those systems are composed of a heterogeneous set of sensors to observe the environment. In centralised systems the observed data will be conveyed to the control room to process the data. Human operators are supposed to give a semantic interpretation of the observed data. They are searching for suspicious or unwanted behaviour. The increase of surveillance sensors in the military domain requires a huge amount of human operators which is far beyond available resources. Automated systems are needed to give a context sensitive semantic interpretation of the observed kinematic data. As a proof of concept two automatic surveillance projects will be discussed in this paper. The first project is about a centralised system based on the AISAutomated Identification System which will be used to monitor ship movements automatically. The second project is about a decentralised system composed of a network of cameras installed at a military area. There is a need for a surveillance system along the coast of Europe. There is an increase of illegal drugs transport from the open sea, intrusion of boat refuges, illegal fishing, pollution of the sea by illegal chemical and oil pollution by ships. An automated sensor system is needed to detect illegal intruders and suspicious ship movements. Vessels fitted with AIS transceivers and transponders can be tracked by AIS base stations located along coast lines or, when out of range of terrestrial networks, through a growing number of satellites that are fitted with special AIS receivers. AIS data include a unique identifier of a vessel and kinematic data such as its position, course and speed. The proposed system enables identification, and tracking of vessels and to detect unwanted or illegal behaviour of ship movements. If ships violate traffic rules, enter forbidden areas or approach a critical infrastructure an alert will be generated automatically in the control room. Human operators start an emergency procedure. The second project is about a network of cameras installed at a military area. The area is monitored by multiple cameras with non-overlapping field of views monitored by human operators. We developed an automated surveillance system. At the entrance gate the identity of visitors will be checked by a face recognition system. In case of intruders, unwanted behaviour, trouble makers the emotional state of the visitor will be assessed by an analysis of facial expressions using the Active Appearance model. If unwanted behaviour is detected an alert is send the control room. Also license place of cars will be recognized using a system based on Neocognitron Neural Networks. Moving objects as persons and vehicles will be detected, localized and tracked. Kinematic parameters are extracted and a semantic interpretation of their behaviour is automatically generated using a rule based system and Bayesian networks. Cars violating the traffic rules or passing speed limits or entering forbidden areas or stopping/parking at forbidden places will be detected. A prototype of a system has been developed which is able to monitor the area 24 hours a day, 7 days a week.

  • [2] Rafei M., Sorkhabi S. E., Mosavi M. R. (Iran)
    Multi-objective optimization by means of multi-dimensional MLP neural networks,31-56

    First page   Full text     DOI: 10.14311/NNW.2014.24.002

    Abstract: In this paper, a multi-layer perceptron (MLP) neural network (NN) is put forward as an efficient tool for performing two tasks: 1) optimization of multi-objective problems and 2) solving a non-linear system of equations. In both cases, mathematical functions which are continuous and partially bounded are involved. Previously, these two tasks were performed by recurrent neural networks and also strong algorithms like evolutionary ones. In this study, multi-dimensional structure in the output layer of the MLP-NN, as an innovative method, is utilized to implicitly optimize the multivariate functions under the network energy optimization mechanism. To this end, the activation functions in the output layer are replaced with the multivariate functions intended to be optimized. The effective training parameters in the global search are surveyed. Also, it is demonstrated that the MLP-NN with proper dynamic learning rate is able to find globally optimal solutions. Finally, the efficiency of the MLP-NN in both aspects of speed and power is investigated by some well-known experimental examples. In some of these examples, the proposed method gives explicitly better globally optimal solutions compared to that of the other references and also shows completely satisfactory results in other experiments.

  • [3] Mocková D., Rybičková A. (CZ)
    Application of genetic algorithms to vehicle routing problem, 57-78

    First page   Full text     DOI: 10.14311/NNW.2014.24.003

    Abstract: Distribution of the goods from a producer to a customer is one of the most important tasks of transportation. This paper focuses on the usage of genetic algorithms (GA) for optimizing problems in transportation, namely vehicle routing problem (VRP). VRP falls in the field of NP-hard problems, which cannot be solved in polynomial time. The problem was solved using genetic algorithm with two types of crossover, both including and leaving-out elitism, setting variable parameters of crossover and mutation probability, as well as prevention of creating invalid individuals. The algorithm was programmed in Matlab, tested on real world problem of spare parts distribution for garages, while the results were compared with another heuristic method (Clarke-Wright method). Genetic algorithm provided a better solution than the heuristic Clarke-Wright method.

  • [4] Mehmet Serhat Odabas, Kadir Ersin Temizel, Omer Caliskan, Nurettin Senyer, Gokhan Kayhan, Erhan Ergun (Turkey)
    Determination of reflectance values of hypericum's leaves under stress conditions using adaptive network based fuzzy inference system, 79-87

    First page   Full text     DOI: 10.14311/NNW.2014.24.004

    Abstract: The effects of water stress and salt levels on hypericum's leaves were examined on greenhouse-grown plants of Hypericum perforatum L. by spectral reflectance. Salt levels and irrigation levels were applied 0, 1, 2.5 and 4 deci Siemens per meter (dS/m), 80%, 100% and 120% respectively. Adaptive Network based Fuzzy Inference System (ANFIS) was performed to estimate the effects of water stress and salt levels on spectral reflectance. As a result of ANFIS, it was found that there was close relationship between actual and predicted reflectance values in Hypericum perforatum L. leaves. Performance of ANFIS was examined under different numbers of epoch and rules. On the other hand, RMSE, correlation and analysis time values were found as outputs. Correlation was 99%. The estimation of optimal ANFIS model was determined in 3*3*3 number of rules with 400 epochs.

  • [5] Jinbao Yao, Baozhen Yao, Yuwei Du, Yonglei Jiang (China)
    Train-induced vibration prediction in multi-story buildings using support vector machine

    First page   Full text     DOI: 10.14311/NNW.2014.24.005

    Abstract: Train-induced vibration prediction in multi-story buildings can effectively provide the effect of vibrations on buildings. With the results of prediction, the corresponding measures can be used to reduce the influence of the vibrations. To accurately predict the vibrations induced by train in multi-story buildings, support vector machine (SVM) is used in this paper. Since the parameters in SVM are very vital for the prediction accuracy, shuffled frog-leaping algorithm (SFLA) is used to optimize the parameters for SVM. The proposed model is evaluated with the data from field experiments. The results show SFLA can effectively provide better parameter values for SVM and the SVM models outperform a better performance than artificial neural network (ANN) for train-induced vibration prediction.

  • [6] Peijuan Li, Yueying Wang, Pingfang Zhou, Quanbao Wang, Dengping Duan (China)
    New results on stability of stochastic neural networks with Markovian switching and mode-dependent time-varying delays, 89-102

    First page   Full text     DOI: 10.14311/NNW.2014.24.006

    Abstract: Abstract: This paper is concerned with the problem of exponential stability for a class of stochastic neural networks with Markovian switching and mode-dependent interval time-varying delays. A novel Lyapunov-Krasovskii functional is introduced with the idea of delay-partitioning, and a new exponential stability criterion is derived based on the new functional and free-weighting matrix method. This new criterion proves to be less conservative than the most existing results. Numerical examples are presented to illustrate the effectiveness of the proposed method.

  • [7] Contents volume 24 (2014), 671-674.
  • [8] Author's index volume 24 (2014), 675-677.