Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057998
I. Reljin, B. Reljin, G. Jovanović
The climate data are In the form of spatial-temporal fields. The most popular method for analyzing such signals is the empirical orthogonal functions (EOF) method. The method is based on the eigenvectors of the spatial cross-covariance matrix of a meteorological field. The EOF method, being linear, is optimal for feature extraction if the data are well characterized by a set of orthogonal structures or functions. Since the dynamics of climate are nonlinear the EOF may become inefficient. Several nonlinear methods for analyzing such fields are known. Here, the nonlinear analysis by using a neural network of the self-organizing map (SOM) structure is applied on the precipitation and the temperature data observed in the region of Yugoslavia.
{"title":"Clustering of climate data in Yugoslavia by using the SOM neural network","authors":"I. Reljin, B. Reljin, G. Jovanović","doi":"10.1109/NEUREL.2002.1057998","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057998","url":null,"abstract":"The climate data are In the form of spatial-temporal fields. The most popular method for analyzing such signals is the empirical orthogonal functions (EOF) method. The method is based on the eigenvectors of the spatial cross-covariance matrix of a meteorological field. The EOF method, being linear, is optimal for feature extraction if the data are well characterized by a set of orthogonal structures or functions. Since the dynamics of climate are nonlinear the EOF may become inefficient. Several nonlinear methods for analyzing such fields are known. Here, the nonlinear analysis by using a neural network of the self-organizing map (SOM) structure is applied on the precipitation and the temperature data observed in the region of Yugoslavia.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132162435","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057976
A. Rodic, D. Katić, M. Vukobratovic
In this paper, a new concept of the advanced integrated vehicle controller with a 4-wheel control system (ADIVEC-4WCS), to provide an automatic system guidance, is presented. The supplementary neuro-compensator is proposed to ensure a control system robustness and better controller adaptability upon the system uncertainties and model inaccuracies. This neural compensator is a part of integrated active control algorithm based on the centralized dynamic control strategy and full vehicle model. The fast convergence of learning process is achieved using standard back propagation method. The validity and effectiveness of the proposed method based on adaptive capability of neural compensator for a four wheel steering system have been demonstrated by simulation experiments.
{"title":"The neural compensator for advance vehicle controller","authors":"A. Rodic, D. Katić, M. Vukobratovic","doi":"10.1109/NEUREL.2002.1057976","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057976","url":null,"abstract":"In this paper, a new concept of the advanced integrated vehicle controller with a 4-wheel control system (ADIVEC-4WCS), to provide an automatic system guidance, is presented. The supplementary neuro-compensator is proposed to ensure a control system robustness and better controller adaptability upon the system uncertainties and model inaccuracies. This neural compensator is a part of integrated active control algorithm based on the centralized dynamic control strategy and full vehicle model. The fast convergence of learning process is achieved using standard back propagation method. The validity and effectiveness of the proposed method based on adaptive capability of neural compensator for a four wheel steering system have been demonstrated by simulation experiments.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113958811","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057995
V. Markovic, Z. Marinković
Low-noise pHEMT transistors, that have excellent performances at microwave frequencies, can be described by their scattering and noise parameters. In this paper, a pHEMT neural model, based on multilayer perceptron neural networks is proposed. The obtained neural models can predict transistor's signal and noise performances very efficiently and accurately for a broad range of bias conditions in the operating frequency range.
{"title":"Signal and noise neural models of pHEMTs","authors":"V. Markovic, Z. Marinković","doi":"10.1109/NEUREL.2002.1057995","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057995","url":null,"abstract":"Low-noise pHEMT transistors, that have excellent performances at microwave frequencies, can be described by their scattering and noise parameters. In this paper, a pHEMT neural model, based on multilayer perceptron neural networks is proposed. The obtained neural models can predict transistor's signal and noise performances very efficiently and accurately for a broad range of bias conditions in the operating frequency range.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130210725","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057991
C. Vasios, G. Matsopoulos, K. Nikita, N. Uzunoğlu, C. Papageorgiou
In this paper a decision support system (DSS) for the classification of patients on their collected event related potentials (ERPs) is proposed. The DSS consists of two levels: the feature extraction level and the classification level. The feature extraction level comprises the implementation of the multivariate autoregressive model in conjunction with a global optimization method, for the selection of optimum features from ERPs. The classification level is implemented with a single three-layer neural network, trained with the backpropagation algorithm and classifies the data into two classes: patients and control subjects. The DSS has been thoroughly tested to a number of patient data (OCD, FES, depressives and drug users), resulting successful classification up to 100%.
{"title":"A decision support system for the classification of event-related potentials","authors":"C. Vasios, G. Matsopoulos, K. Nikita, N. Uzunoğlu, C. Papageorgiou","doi":"10.1109/NEUREL.2002.1057991","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057991","url":null,"abstract":"In this paper a decision support system (DSS) for the classification of patients on their collected event related potentials (ERPs) is proposed. The DSS consists of two levels: the feature extraction level and the classification level. The feature extraction level comprises the implementation of the multivariate autoregressive model in conjunction with a global optimization method, for the selection of optimum features from ERPs. The classification level is implemented with a single three-layer neural network, trained with the backpropagation algorithm and classifies the data into two classes: patients and control subjects. The DSS has been thoroughly tested to a number of patient data (OCD, FES, depressives and drug users), resulting successful classification up to 100%.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132621452","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057982
I. Krstić, B. Reljin, P. Kostic, D. Kandic
In the most general case the finding of the shear stress distribution on the cross section of prismatic bar subjected to torsion is a specific problem that can be solved in two steps. The first of them consists in finding the so-called stress function, and the second one in finding the shear stresses on the basis of the formerly found stress function. The stress function is the solution of Poisson's partial differential equation for given conditions of unambiguity that in the elasticity theory describes the torsion of prismatic bars in terms of stresses. Modeling by means of electrical networks is one of a few possible ways to find the stress function. This paper describes how Chua and Yang's cellular neural networks can be used as an analogous model to find the stress function of a twisted prismatic bar, which serves to calculate the shear stress distribution. Effectiveness of the presented method is illustrated by the solutions of two problems. The method can be applied in mechanical and civil engineering.
{"title":"Application of cellular neural networks in stress analysis of prismatic bars subjected to torsion","authors":"I. Krstić, B. Reljin, P. Kostic, D. Kandic","doi":"10.1109/NEUREL.2002.1057982","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057982","url":null,"abstract":"In the most general case the finding of the shear stress distribution on the cross section of prismatic bar subjected to torsion is a specific problem that can be solved in two steps. The first of them consists in finding the so-called stress function, and the second one in finding the shear stresses on the basis of the formerly found stress function. The stress function is the solution of Poisson's partial differential equation for given conditions of unambiguity that in the elasticity theory describes the torsion of prismatic bars in terms of stresses. Modeling by means of electrical networks is one of a few possible ways to find the stress function. This paper describes how Chua and Yang's cellular neural networks can be used as an analogous model to find the stress function of a twisted prismatic bar, which serves to calculate the shear stress distribution. Effectiveness of the presented method is illustrated by the solutions of two problems. The method can be applied in mechanical and civil engineering.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116795022","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057992
K. Jovanova-Nešić, M. Eric-Jovicic, M. Popovic, N. Popovic, L. Rakić, N. H. Spector
In a previous paper, the authors (1997) have described the effect of Ca/sup 2+/-antagonist verapamil on Na,K-ATPase in experimental model of Alzheimer's disease (AD). The present paper is concerned with the effect of magnetic stimulation of pineal complex on Na,K-ATPase activity in the same experimental model of AD. Since accumulating data indicate that free radicals mediate injury and death of neurons in AD, and because magnetic fields (MFs) can alter free radicals reactions, we tested the hypothesis that stationary MFs mediates ion homeostasis through membrane Na,K-ATPase activity. Results are presented as Vmax/Km parameters on erythrocyte membranes in peripheral blood of rats with lesioned nucleus basalis magnocellularis. Bilateral electrolytic or by kainic acid induced lesions of NBM induce significant decrease of Vmax/Km activity on erythrocyte membranes obtained by cardiac function. Stimulation of pineal complex of the brain more than ten days, by magnetic beads (600-Gauss flux density) fixed on the skull upon pineal gland, significantly increase impaired by lesions of NBM, Na,K-ATPase activity. Results are presented as Vmax/Km parameters on erythrocyte membranes in peripheral blood of rats with lesioned NBM of the basal forebrain bundle. These results confirm the hypothesis that altered ion homeostasis disturbed by neuro-degenerations play an essential role in pathogenesis of experimental AD and that magnetic stimulation of the pineal complex might successfully restore disturbed by neuronal death Na,K-ATPase activity.
在之前的论文中，作者(1997)描述了Ca/sup 2+/-拮抗剂维拉帕米对阿尔茨海默病(AD)实验模型中Na, k - atp酶的影响。本文研究了磁场刺激松果体复合体对AD实验模型中Na, k - atp酶活性的影响。由于积累的数据表明自由基介导AD中神经元的损伤和死亡，并且由于磁场(MFs)可以改变自由基反应，我们验证了静止的MFs通过膜Na, k - atp酶活性介导离子稳态的假设。结果为大细胞基底核损伤大鼠外周血红细胞膜Vmax/Km参数。双侧电解或kainic酸诱导的NBM病变导致心功能获得的红细胞膜Vmax/Km活性显著降低。刺激大脑松果体复合体十天以上，通过磁珠(600高斯磁密度)固定在颅骨上的松果体上，明显增加受损病变的NBM、Na、k - atp酶活性。结果为基底前脑束NBM损伤大鼠外周血红细胞膜Vmax/Km参数。这些结果证实了由神经变性引起的离子稳态改变在实验性AD发病机制中起重要作用的假设，并且松果体复合体的磁刺激可能成功恢复受神经元死亡干扰的Na, k - atp酶活性。
{"title":"Effect of magnetic stimulation of pineal complex of the brain on Na,K-ATPase in experimental Alzheimer's disease","authors":"K. Jovanova-Nešić, M. Eric-Jovicic, M. Popovic, N. Popovic, L. Rakić, N. H. Spector","doi":"10.1109/NEUREL.2002.1057992","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057992","url":null,"abstract":"In a previous paper, the authors (1997) have described the effect of Ca/sup 2+/-antagonist verapamil on Na,K-ATPase in experimental model of Alzheimer's disease (AD). The present paper is concerned with the effect of magnetic stimulation of pineal complex on Na,K-ATPase activity in the same experimental model of AD. Since accumulating data indicate that free radicals mediate injury and death of neurons in AD, and because magnetic fields (MFs) can alter free radicals reactions, we tested the hypothesis that stationary MFs mediates ion homeostasis through membrane Na,K-ATPase activity. Results are presented as Vmax/Km parameters on erythrocyte membranes in peripheral blood of rats with lesioned nucleus basalis magnocellularis. Bilateral electrolytic or by kainic acid induced lesions of NBM induce significant decrease of Vmax/Km activity on erythrocyte membranes obtained by cardiac function. Stimulation of pineal complex of the brain more than ten days, by magnetic beads (600-Gauss flux density) fixed on the skull upon pineal gland, significantly increase impaired by lesions of NBM, Na,K-ATPase activity. Results are presented as Vmax/Km parameters on erythrocyte membranes in peripheral blood of rats with lesioned NBM of the basal forebrain bundle. These results confirm the hypothesis that altered ion homeostasis disturbed by neuro-degenerations play an essential role in pathogenesis of experimental AD and that magnetic stimulation of the pineal complex might successfully restore disturbed by neuronal death Na,K-ATPase activity.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114693844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057961
B. Todorovic, M. Stankovic, C. Moraga
This paper addresses the problem of continuous adaptation of neural networks in a non-stationary environment. We have applied the extended Kalman filter to the parameter, state and structure estimation of a recurrent radial basis function network. The architecture of the recurrent radial basis function network implements a nonlinear autoregressive model with exogenous inputs. Statistical criteria for structure adaptation (growing and pruning of hidden units and connections of the network) were derived using statistics estimated by the Kalman filter. The proposed algorithm is applied to non-stationary dynamic system modeling.
{"title":"Modeling non-stationary dynamic system using recurrent radial basis function networks","authors":"B. Todorovic, M. Stankovic, C. Moraga","doi":"10.1109/NEUREL.2002.1057961","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057961","url":null,"abstract":"This paper addresses the problem of continuous adaptation of neural networks in a non-stationary environment. We have applied the extended Kalman filter to the parameter, state and structure estimation of a recurrent radial basis function network. The architecture of the recurrent radial basis function network implements a nonlinear autoregressive model with exogenous inputs. Statistical criteria for structure adaptation (growing and pruning of hidden units and connections of the network) were derived using statistics estimated by the Kalman filter. The proposed algorithm is applied to non-stationary dynamic system modeling.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126002168","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057969
I. Gavat, C. Dumitru, G. Costache
In the paper is presented our work concerning speech processing by neural networks. A combination of Kohonen maps and multilayer perceptrons is applied in a word spotting task. A hierarchical segmentation procedure for continuous speech is realized with multilayer perceptrons, Kohonen maps and radial basis function networks. A neuro-statistical structure for isolated word recognition and a neuro-fuzzy hybrid for vowel recognition are analyzed.
{"title":"Applications of neural networks in speech processing for Romanian language","authors":"I. Gavat, C. Dumitru, G. Costache","doi":"10.1109/NEUREL.2002.1057969","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057969","url":null,"abstract":"In the paper is presented our work concerning speech processing by neural networks. A combination of Kohonen maps and multilayer perceptrons is applied in a word spotting task. A hierarchical segmentation procedure for continuous speech is realized with multilayer perceptrons, Kohonen maps and radial basis function networks. A neuro-statistical structure for isolated word recognition and a neuro-fuzzy hybrid for vowel recognition are analyzed.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129966365","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057966
S. Vrettos, A. Stafylopatis
This work proposes a fuzzy logic framework suitable for web page filtering. Web page classifiers are trained off-line using the directory structure of the Open Directory Project (http://dmoz.org/) and are available to the user through an appropriate interface. These classifiers are considered as fuzzy membership functions which determine the membership degree of a web page to each class. The user selects a number of classes and formulates logical rules combining the classifiers. Fuzzy logic operators are used in order to filter the results of a query according to the specified rule providing different views (orderings) of the search results to the user.
{"title":"A fuzzy logic framework for web page filtering","authors":"S. Vrettos, A. Stafylopatis","doi":"10.1109/NEUREL.2002.1057966","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057966","url":null,"abstract":"This work proposes a fuzzy logic framework suitable for web page filtering. Web page classifiers are trained off-line using the directory structure of the Open Directory Project (http://dmoz.org/) and are available to the user through an appropriate interface. These classifiers are considered as fuzzy membership functions which determine the membership degree of a web page to each class. The user selects a number of classes and formulates logical rules combining the classifiers. Fuzzy logic operators are used in order to filter the results of a query according to the specified rule providing different views (orderings) of the search results to the user.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124270782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-12-10DOI: 10.1109/NEUREL.2002.1057977
P. Radonja, S. Stankovic
In this paper, we attempt, using an artificial intelligence method based on neural networks, to obtain a model of a nonlinear process from observed datasets. In the first part of the paper, six different processes are analyzed on the basis of small data sets and divided into two groups. After that, the corresponding data-based models are generated for the obtained two groups of measured data sets. In the following, the proposed models are tested on two new data sets.
{"title":"Neural network models based on small data sets","authors":"P. Radonja, S. Stankovic","doi":"10.1109/NEUREL.2002.1057977","DOIUrl":"https://doi.org/10.1109/NEUREL.2002.1057977","url":null,"abstract":"In this paper, we attempt, using an artificial intelligence method based on neural networks, to obtain a model of a nonlinear process from observed datasets. In the first part of the paper, six different processes are analyzed on the basis of small data sets and divided into two groups. After that, the corresponding data-based models are generated for the obtained two groups of measured data sets. In the following, the proposed models are tested on two new data sets.","PeriodicalId":347066,"journal":{"name":"6th Seminar on Neural Network Applications in Electrical Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116950594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：