Pełnotekstowe zasoby PLDML oraz innych baz dziedzinowych są już dostępne w nowej Bibliotece Nauki.
Zapraszamy na https://bibliotekanauki.pl
Preferencje help
Polish version English version Język
Widoczny [Schowaj] Abstrakt
Liczba wyników

Znaleziono wyników: 26

Liczba wyników na stronie
first rewind previous Strona / 2 next fast forward last

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych:  neural networks
help Sortuj według:

help Ogranicz wyniki do:
first rewind previous Strona / 2 next fast forward last
1
Content available remote

Evolving co-adapted subcomponents in assembler encoding

100%
Praczyk T.
International Journal of Applied Mathematics and Computer Science
|
2007
|
tom 17
|
nr 4
549-563
EN
The paper presents a new Artificial Neural Network (ANN) encoding method called Assembler Encoding (AE). It assumes that the ANN is encoded in the form of a program (Assembler Encoding Program, AEP) of a linear organization and of a structure similar to the structure of a simple assembler program. The task of the AEP is to create a Connectivity Matrix (CM) which can be transformed into the ANN of any architecture. To create AEPs, and in consequence ANNs, genetic algorithms (GAs) are used. In addition to the outline of AE, the paper also presents a new AEP encoding method, i.e., the method used to represent the AEP in the form of a chromosome or a set of chromosomes. The proposed method assumes the evolution of individual components of AEPs, i.e., operations and data, in separate populations. To test the method, experiments in two areas were carried out, i.e., in optimization and in a predator-prey problem. In the first case, the task of AE was to create matrices which constituted a solution to the optimization problem. In the second case, AE was responsible for constructing neural controllers used to control artificial predators whose task was to capture a fast-moving prey.
2
Content available remote

Neural network-based NARX models in non-linear adaptive control

100%
Dzieliński A.
International Journal of Applied Mathematics and Computer Science
|
2002
|
tom 12
|
nr 2
235-240
EN
The applicability of approximate NARX models of non-linear dynamic systems is discussed. The models are obtained by a new version of Fourier analysis-based neural network also described in the paper. This constitutes a reformulation of a known method in a recursive manner, i.e. adapted to account for incoming data on-line. The method allows us to obtain an approximate model of the non-linear system. The estimation of the influence of the modelling error on the discrepancy between the model and real system outputs is given. Possible applications of this approach to the design of BIBO stable closed-loop control are proposed.
3
Content available remote

The UD RLS algorithm for training feedforward neural networks

100%
Bilski J.
International Journal of Applied Mathematics and Computer Science
|
2005
|
tom 15
|
nr 1
115-123
EN
A new algorithm for training feedforward multilayer neural networks is proposed. It is based on recursive least squares procedures and U-D factorization, which is a well-known technique in filter theory. It will be shown that due to the U-D factorization method, our algorithm requires fewer computations than the classical RLS applied to feedforward multilayer neural network training.
4
Content available remote

Ensemble neural network approach for accurate load forecasting in a power system

100%
Siwek K., Osowski S., Szupiluk R.
International Journal of Applied Mathematics and Computer Science
|
2009
|
tom 19
|
nr 2
303-315
EN
The paper presents an improved method for 1-24 hours load forecasting in the power system, integrating and combining different neural forecasting results by an ensemble system. We will integrate the results of partial predictions made by three solutions, out of which one relies on a multilayer perceptron and two others on self-organizing networks of the competitive type. As the expert system we will apply different integration methods: simple averaging, SVD based weighted averaging, principal component analysis and blind source separation. The results of numerical experiments, concerning forecasting the hourly load for the next 24 hours of the Polish power system, will be presented and discussed. We will compare the performance of different ensemble methods on the basis of the mean absolute percentage error, mean squared error and maximum percentage error. They show a significant improvement of the proposed ensemble method in comparison to the individual results of prediction. The comparison of our work with the results of other papers for the same data proves the superiority of our approach.
5
Content available remote

Advances in model-based fault diagnosis with evolutionary algorithms and neural networks

100%
Witczak M.
International Journal of Applied Mathematics and Computer Science
|
2006
|
tom 16
|
nr 1
85-99
EN
Challenging design problems arise regularly in modern fault diagnosis systems. Unfortunately, the classical analytical techniques often cannot provide acceptable solutions to such difficult tasks. This explains why soft computing techniques such as evolutionary algorithms and neural networks become more and more popular in industrial applications of fault diagnosis. The main objective of this paper is to present recent developments regarding the application of evolutionary algorithms and neural networks to fault diagnosis. In particular, a brief introduction to these computational intelligence paradigms is presented, and then a review of their fault detection and isolation applications is performed. Close attention is paid to techniques that integrate the classical and soft computing methods. A selected group of them is carefully described in the paper. The performance of the presented approaches is illustrated with the use of the DAMADICS fault detection benchmark that deals with a valve actuator.
6
Artykuł dostępny w postaci pełnego tekstu - kliknij by otworzyć plik
Content available

A connectionist computational method for face recognition

100%
Pujol F., Mora H., Girona-Selva J.
International Journal of Applied Mathematics and Computer Science
|
2016
|
tom 26
|
nr 2
451-465
EN
In this work, a modified version of the elastic bunch graph matching (EBGM) algorithm for face recognition is introduced. First, faces are detected by using a fuzzy skin detector based on the RGB color space. Then, the fiducial points for the facial graph are extracted automatically by adjusting a grid of points to the result of an edge detector. After that, the position of the nodes, their relation with their neighbors and their Gabor jets are calculated in order to obtain the feature vector defining each face. A self-organizing map (SOM) framework is shown afterwards. Thus, the calculation of the winning neuron and the recognition process are performed by using a similarity function that takes into account both the geometric and texture information of the facial graph. The set of experiments carried out for our SOM-EBGM method shows the accuracy of our proposal when compared with other state-of the-art methods.
7
Content available remote

A new approach to image reconstruction from projections using a recurrent neural network

100%
Cierniak R.
International Journal of Applied Mathematics and Computer Science
|
2008
|
tom 18
|
nr 2
147-157
EN
A new neural network approach to image reconstruction from projections considering the parallel geometry of the scanner is presented. To solve this key problem in computed tomography, a special recurrent neural network is proposed. The reconstruction process is performed during the minimization of the energy function in this network. The performed computer simulations show that the neural network reconstruction algorithm designed to work in this way outperforms conventional methods in the obtained image quality.
8
Content available remote

Improving the generalization ability of neuro-fuzzy systems by ε-insensitive learning

88%
Łęski J.
International Journal of Applied Mathematics and Computer Science
|
2002
|
tom 12
|
nr 3
437-447
EN
A new learning method tolerant of imprecision is introduced and used in neuro-fuzzy modelling. The proposed method makes it possible to dispose of an intrinsic inconsistency of neuro-fuzzy modelling, where zero-tolerance learning is used to obtain a fuzzy model tolerant of imprecision. This new method can be called ε-insensitive learning, where, in order to fit the fuzzy model to real data, the ε-insensitive loss function is used. ε-insensitive learning leads to a model with minimal Vapnik-Chervonenkis dimension, which results in an improved generalization ability of this system. Another advantage of the proposed method is its robustness against outliers. This paper introduces two approaches to solving ε-insensitive learning problem. The first approach leads to a quadratic programming problem with bound constraints and one linear equality constraint. The second approach leads to a problem of solving a system of linear inequalities. Two computationally efficient numerical methods for ε-insensitive learning are proposed. Finally, examples are given to demonstrate the validity of the introduced methods.
9
Content available remote

Fuzzy and neural control of an induction motor

88%
Denai M. A., Attia S. A.
International Journal of Applied Mathematics and Computer Science
|
2002
|
tom 12
|
nr 2
221-233
EN
This paper presents some design approaches to hybrid control systems combining conventional control techniques with fuzzy logic and neural networks. Such a mixed implementation leads to a more effective control design with improved system performance and robustness. While conventional control allows different design objectives such as steady state and transient characteristics of the closed loop system to be specified, fuzzy logic and neural networks are integrated to overcome the problems with uncertainties in the plant parameters and structure encountered in the classical model-based design. Induction motors are characterised by complex, highly non-linear and time-varying dynamics and inaccessibility of some states and outputs for measurements, and hence can be considered as a challenging engineering problem. The advent of vector control techniques has partially solved induction motor control problems, because they are sensitive to drive parameter variations and performance may deteriorate if conventional controllers are used. Fuzzy logic and neural network-based controllers are considered as potential candidates for such an application. Three control approaches are developed and applied to adjust the speed of the drive system. The first control design combines the variable structure theory with the fuzzy logic concept. In the second approach neural networks are used in an internal model control structure. Finally, a fuzzy state feedback controller is developed based on the pole placement technique. A simulation study of these methods is presented. The effectiveness of these controllers is demonstrated for different operating conditions of the drive system.
10
Artykuł dostępny w postaci pełnego tekstu - kliknij by otworzyć plik
Content available

Neural networks as a tool for georadar data processing

88%
Szymczyk P., Tomecka-Suchoń S., Szymczyk M.
International Journal of Applied Mathematics and Computer Science
|
2015
|
tom 25
|
nr 4
955-960
EN
In this article a new neural network based method for automatic classification of ground penetrating radar (GPR) traces is proposed. The presented approach is based on a new representation of GPR signals by polynomials approximation. The coefficients of the polynomial (the feature vector) are neural network inputs for automatic classification of a special kind of geologic structure-a sinkhole. The analysis and results show that the classifier can effectively distinguish sinkholes from other geologic structures.
11
Content available remote

Nonlinear predictive control based on neural multi-models

88%
Ławryńczuk M., Tatjewski P.
International Journal of Applied Mathematics and Computer Science
|
2010
|
tom 20
|
nr 1
7-21
EN
This paper discusses neural multi-models based on Multi Layer Perceptron (MLP) networks and a computationally efficient nonlinear Model Predictive Control (MPC) algorithm which uses such models. Thanks to the nature of the model it calculates future predictions without using previous predictions. This means that, unlike the classical Nonlinear Auto Regressive with eXternal input (NARX) model, the multi-model is not used recurrently in MPC, and the prediction error is not propagated. In order to avoid nonlinear optimisation, in the discussed suboptimal MPC algorithm the neural multi-model is linearised on-line and, as a result, the future control policy is found by solving of a quadratic programming problem.
12
Content available remote

Convergence analysis for principal component flows

88%
Yoshizawa S., Helmke U., Starkov K.
International Journal of Applied Mathematics and Computer Science
|
2001
|
tom 11
|
nr 1
223-236
EN
A common framework for analyzing the global convergence of several flows for principal component analysis is developed. It is shown that flows proposed by Brockett, Oja, Xu and others are all gradient flows and the global convergence of these flows to single equilibrium points is established. The signature of the Hessian at each critical point is determined.
13
Content available remote

Time-varying time-delay estimation for nonlinear systems using neural networks

88%
Tan Y.
International Journal of Applied Mathematics and Computer Science
|
2004
|
tom 14
|
nr 1
63-68
EN
Nonlinear dynamic processes with time-varying time delays can often be encountered in industry. Time-delay estimation for nonlinear dynamic systems with time-varying time delays is an important issue for system identification. In order to estimate the dynamics of a process, a dynamic neural network with an external recurrent structure is applied in the modeling procedure. In the case where a delay is time varying, a useful way is to develop on-line time-delay estimation mechanisms to track the time-delay variation. In this paper, two schemes called direct and indirect time-delay estimators are proposed. The indirect time-delay estimator considers the procedure of time-delay estimation as a nonlinear programming problem. On the other hand, the direct time-delay estimation scheme applies a neural network to construct a time-delay estimator to track the time-varying time-delay. Finally, a numerical example is considered for testing the proposed methods.
14
Content available remote

A family of model predictive control algorithms with artificial neural networks

88%
Ławryńczuk M.
International Journal of Applied Mathematics and Computer Science
|
2007
|
tom 17
|
nr 2
217-232
EN
This paper details nonlinear Model-based Predictive Control (MPC) algorithms for MIMO processes modelled by means of neural networks of a feedforward structure. Two general MPC techniques are considered: the one with Nonlinear Optimisation (MPC-NO) and the one with Nonlinear Prediction and Linearisation (MPC-NPL). In the first case a nonlinear optimisation problem is solved in real time on-line. In order to reduce the computational burden, in the second case a neural model of the process is used on-line to determine local linearisation and a nonlinear free trajectory. Single-point and multi-point linearisation methods are discussed. The MPC-NPL structure is far more reliable and less computationally demanding in comparison with the MPC-NO one because it solves a quadratic programming problem, which can be done efficiently within a foreseeable time frame. At the same time, closed-loop performance of both algorithm classes is similar. Finally, a hybrid MPC algorithm with Nonlinear Prediction, Linearisation and Nonlinear optimisation (MPC-NPL-NO) is discussed.
15
Content available remote

Input constraints handling in an MPC/feedback linearization scheme

88%
Deng J., Becerra V., Stobart R.
International Journal of Applied Mathematics and Computer Science
|
2009
|
tom 19
|
nr 2
219-232
EN
The combination of model predictive control based on linear models (MPC) with feedback linearization (FL) has attracted interest for a number of years, giving rise to MPC+FL control schemes. An important advantage of such schemes is that feedback linearizable plants can be controlled with a linear predictive controller with a fixed model. Handling input constraints within such schemes is difficult since simple bound contraints on the input become state dependent because of the nonlinear transformation introduced by feedback linearization. This paper introduces a technique for handling input constraints within a real time MPC/FL scheme, where the plant model employed is a class of dynamic neural networks. The technique is based on a simple affine transformation of the feasible area. A simulated case study is presented to illustrate the use and benefits of the technique.
16
Content available remote

Nonparametric statistical analysis for multiple comparison of machine learning regression algorithms

88%
Trawiński B., Smętek M., Telec Z., Lasota T.
International Journal of Applied Mathematics and Computer Science
|
2012
|
tom 22
|
nr 4
867-881
EN
In the paper we present some guidelines for the application of nonparametric statistical tests and post-hoc procedures devised to perform multiple comparisons of machine learning algorithms. We emphasize that it is necessary to distinguish between pairwise and multiple comparison tests. We show that the pairwise Wilcoxon test, when employed to multiple comparisons, will lead to overoptimistic conclusions. We carry out intensive normality examination employing ten different tests showing that the output of machine learning algorithms for regression problems does not satisfy normality requirements. We conduct experiments on nonparametric statistical tests and post-hoc procedures designed for multiple 1 × N and N × N comparisons with six different neural regression algorithms over 29 benchmark regression data sets. Our investigation proves the usefulness and strength of multiple comparison statistical procedures to analyse and select machine learning algorithms.
17
Content available remote

Efficient nonlinear predictive control based on structured neural models

88%
Ławryńczuk M.
International Journal of Applied Mathematics and Computer Science
|
2009
|
tom 19
|
nr 2
233-246
EN
This paper describes structured neural models and a computationally efficient (suboptimal) nonlinear Model Predictive Control (MPC) algorithm based on such models. The structured neural model has the ability to make future predictions of the process without being used recursively. Thanks to the nature of the model, the prediction error is not propagated. This is particularly important in the case of noise and underparameterisation. Structured models have much better long-range prediction accuracy than the corresponding classical Nonlinear Auto Regressive with eXternal input (NARX) models. The described suboptimal MPC algorithm needs solving on-line only a quadratic programming problem. Nevertheless, it gives closed-loop control performance similar to that obtained in fully-fledged nonlinear MPC, which hinges on online nonconvex optimisation. In order to demonstrate the advantages of structured models as well as the accuracy of the suboptimal MPC algorithm, a polymerisation reactor is studied.
18
Content available remote

Soft computing in modelbased predictive control footnotemark

88%
Tatjewski P., Ławrynczuk M.
International Journal of Applied Mathematics and Computer Science
|
2006
|
tom 16
|
nr 1
7-26
EN
The application of fuzzy reasoning techniques and neural network structures to model-based predictive control (MPC) is studied. First, basic structures of MPC algorithms are reviewed. Then, applications of fuzzy systems of the Takagi-Sugeno type in explicit and numerical nonlinear MPC algorithms are presented. Next, many techniques using neural network modeling to improve structural or computational properties of MPC algorithms are presented and discussed, from a neural network model of a process in standard MPC structures to modeling parts or entire MPC controllers with neural networks. Finally, a simulation example and conclusions are given.
19
Artykuł dostępny w postaci pełnego tekstu - kliknij by otworzyć plik
Content available

Data mining methods for prediction of air pollution

88%
Siwek K., Osowski S.
International Journal of Applied Mathematics and Computer Science
|
2016
|
tom 26
|
nr 2
467-478
EN
The paper discusses methods of data mining for prediction of air pollution. Two tasks in such a problem are important: generation and selection of the prognostic features, and the final prognostic system of the pollution for the next day. An advanced set of features, created on the basis of the atmospheric parameters, is proposed. This set is subject to analysis and selection of the most important features from the prediction point of view. Two methods of feature selection are compared. One applies a genetic algorithm (a global approach), and the other-a linear method of stepwise fit (a locally optimized approach). On the basis of such analysis, two sets of the most predictive features are selected. These sets take part in prediction of the atmospheric pollutants PM10, SO2, NO2 and O3. Two approaches to prediction are compared. In the first one, the features selected are directly applied to the random forest (RF), which forms an ensemble of decision trees. In the second case, intermediate predictors built on the basis of neural networks (the multilayer perceptron, the radial basis function and the support vector machine) are used. They create an ensemble integrated into the final prognosis. The paper shows that preselection of the most important features, cooperating with an ensemble of predictors, allows increasing the forecasting accuracy of atmospheric pollution in a significant way.
20
Content available remote

Approximation of phenol concentration using novel hybrid computational intelligence methods

75%
Pławiak P., Tadeusiewicz R.
International Journal of Applied Mathematics and Computer Science
|
2014
|
tom 24
|
nr 1
165-181
EN
This paper presents two innovative evolutionary-neural systems based on feed-forward and recurrent neural networks used for quantitative analysis. These systems have been applied for approximation of phenol concentration. Their performance was compared against the conventional methods of artificial intelligence (artificial neural networks, fuzzy logic and genetic algorithms). The proposed systems are a combination of data preprocessing methods, genetic algorithms and the Levenberg-Marquardt (LM) algorithm used for learning feed forward and recurrent neural networks. The initial weights and biases of neural networks chosen by the use of a genetic algorithm are then tuned with an LM algorithm. The evaluation is made on the basis of accuracy and complexity criteria. The main advantage of proposed systems is the elimination of random selection of the network weights and biases, resulting in increased efficiency of the systems.
first rewind previous Strona / 2 next fast forward last
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.