Informacja

Drogi użytkowniku, aplikacja do prawidłowego działania wymaga obsługi JavaScript. Proszę włącz obsługę JavaScript w Twojej przeglądarce.

English (EN)·Polski (PL)·Yкраїнський (UK)
Przejdź do strony domowej biblioteki Centralna Biblioteka Wojskowa Link otwiera się w nowym oknie Logo biblioteki Prolib Integro - strona głównaCentralna Biblioteka Wojskowa

Wyszukujesz frazę "function networks" wg kryterium: Wszystkie pola

  Lista filtrów
Wyrównaj
Wyświetlanie 1-15 z 47
Tytuł:
Prediction of industrial pollution by radial basis function networks
Autorzy:
Djebbri, N.
Rouainia, M.
Powiązania:
https://bibliotekanauki.pl/articles/207579.pdf
Data publikacji:
2018
Wydawca:
Politechnika Wrocławska. Oficyna Wydawnicza Politechniki Wrocławskiej
Tematy:
forecasting
RBF
artificial neural network
pollution
prognozowanie
sztuczna sieć neuronowa
zanieczyszczenie
Opis:
Atmospheric pollution has been receiving a significant interest for several decades since industries cause more and more pollution. Thanks to the development of many prediction techniques, scientists and industries are focusing more on pollution prediction. The aim of this work is to predict the two pollutant concentrations (NOx and CO) in industrial sites by a modified radial basis function (RBF) based neural network. The modification considered the spread parameter h of the activation function in the RBF network. In order to get the best network, the variations of this parameter for three cases were considered. In the first case, only pollutants concentrations variables were used, while in the second one, only the meteorological variables were utilized. In the third case, pollutants' concentrations were connected with meteorological variables. Based on calculation errors, the best model that ensures the best monitoring of pollutants concentration could be identified.
Źródło:
Environment Protection Engineering; 2018, 44, 3; 153-164
0324-8828
Pojawia się w:
Environment Protection Engineering
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Prediction of Impact Resistance Properties of Concrete Using Radial Basis Function Networks
Autorzy:
Yazici, S.
Inan Sezer, G.
Sezer, A.
Powiązania:
https://bibliotekanauki.pl/articles/1031254.pdf
Data publikacji:
2017-09
Wydawca:
Polska Akademia Nauk. Instytut Fizyki PAN
Tematy:
81.70.Bt
79.20.Ap
07.05.Mh
Opis:
This study presents an investigation of the prediction of impact resistance of steel-fiber-reinforced concrete and ordinary concrete specimens. In the experimental part of this study, parameters identifying impact resistance of various concrete mixtures were determined using an impact test machine, in accordance with ACI Committee 544. For this aim, concrete specimens containing three different aggregates (basalt, limestone and natural aggregate) were cured in water at 20°C for 28 days. After curing impact resistance tests were performed on specimens having compressive strength values between 20 and 50 MPa, to determine the blows to initial crack and failure. The specimens were also subjected to splitting tensile strength and ultrasonic pulse velocity tests. Initially, using blows to initial crack and failure, many attempts were made to classify the impact resistance of different types of concrete in terms of the origin of used aggregate, strength properties or ultrasonic pulse velocity, however, this made no sense. The specimens could only be classified in terms of steel fiber presence. Therefore, radial basis function network, which belongs to another kind of unsupervised classifier network, was used to estimate the two above-mentioned impact resistance parameters. In this scope, independent from aggregate origin used in preparation of specimens, compressive strength, splitting tensile strength and ultrasonic pulse velocity of the specimens were used to predict the impact resistance parameters of the concrete specimens. The results revealed that three listed parameters can be used for estimation of blows to formation of initial crack and failure. Scatter plots, root mean square error and absolute value of average residual parameters were used to verify the errors in predictions, which were very low, compared with the uncertainty in test and ambiguity of data in hand.
Źródło:
Acta Physica Polonica A; 2017, 132, 3; 1036-1040
0587-4246
1898-794X
Pojawia się w:
Acta Physica Polonica A
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Stabilising solutions to a class of nonlinear optimal state tracking problems using radial basis function networks
Autorzy:
Ahmida, Z.
Charef, A.
Becerra, V. M.
Powiązania:
https://bibliotekanauki.pl/articles/908523.pdf
Data publikacji:
2005
Wydawca:
Uniwersytet Zielonogórski. Oficyna Wydawnicza
Tematy:
system nieliniowy
sterowanie optymalne
radialna funkcja bazowa
sieć neuronowa
regulacja predykcyjna
sterowanie wyprzedzające
nonlinear systems
optimal control
radial basis functions
neural networks
predictive control
feedforward control
Opis:
A controller architecture for nonlinear systems described by Gaussian RBF neural networks is proposed. The controller is a stabilising solution to a class of nonlinear optimal state tracking problems and consists of a combination of a state feedback stabilising regulator and a feedforward neuro-controller. The state feedback stabilising regulator is computed online by transforming the tracking problem into a more manageable regulation one, which is solved within the framework of a nonlinear predictive control strategy with guaranteed stability. The feedforward neuro-controller has been designed using the concept of inverse mapping. The proposed control scheme is demonstrated on a simulated single-link robotic manipulator.
Źródło:
International Journal of Applied Mathematics and Computer Science; 2005, 15, 3; 369-381
1641-876X
2083-8492
Pojawia się w:
International Journal of Applied Mathematics and Computer Science
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Membership function - ARTMAP neural networks
Autorzy:
Sinčák, P.
Hric, M.
Vaščák, J.
Powiązania:
https://bibliotekanauki.pl/articles/1931570.pdf
Data publikacji:
2003
Wydawca:
Politechnika Gdańska
Tematy:
pattern recognition principles
classifier design
classification accuracy assessment
contingency tables
backpropagation neural networks
fuzzy BP neural networks
ART and ARTMAP neural networks
modular neural networks
neural networks
Opis:
The project deals with the application of computational intelligence (CI) tools for multispectral image classification. Pattern Recognition scheme is a global approach where the classification part is playing an important role to achieve the highest classification accuracy. Multispectral images are data mainly used in remote sensing and this kind of classification is very difficult to assess the accuracy of classification results. There is a feedback problem in adjusting the parts of pattern recognition scheme. Precise classification accuracy assessment is almost impossible to obtain, being an extremely laborious procedure. The paper presents simple neural networks for multispectral image classification, ARTMAP-like neural networks as more sophisticated tools for classification, and a modular approach to achieve the highest classification accuracy of multispectral images. There is a strong link to advances in computer technology, which gives much better conditions for modelling more sophisticated classifiers for multispectral images.
Źródło:
TASK Quarterly. Scientific Bulletin of Academic Computer Centre in Gdansk; 2003, 7, 1; 43-52
1428-6394
Pojawia się w:
TASK Quarterly. Scientific Bulletin of Academic Computer Centre in Gdansk
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
The Analytical and Artificial Intelligence Methods to Investigate the Effects of Aperture Dimension Ratio on Electrical Shielding Effectiveness
Autorzy:
Basyigit, Ibrahim Bahadir
Dogan, Habib
Powiązania:
https://bibliotekanauki.pl/articles/226583.pdf
Data publikacji:
2019
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Tematy:
electromagnetic shielding
electromagnetic
compatibility
apertures
multilayer perceptron
radial basis
function networks
Opis:
This paper presents that the effect of single aperture size of metallic enclosure on electrical shielding effectiveness (ESE) at 0 – 1 GHz frequency range has been investigated by using both Robinson’s analytical formulation and artificial neural networks (ANN) methods that are multilayer perceptron (MLP) networks and a radial basis function neural network (RBFNN). All results including measurement have been compared each other in terms of aperture geometry of metallic enclosure. The geometry of single aperture varies from square to rectangular shape while the open area of aperture is fixed. It has been observed that network structure of MLP 3-40-1 in modeling with ANN modeled with fewer neurons in the sense of overlapping of faults and data and modeled accordingly. In contrast, the RBFNN 3-150-1 is the other detection that the network structure is modeled with more neurons and more. It can be seen from the same network-structured MLP and RBFNN that the MLP modeled better. In this paper, the impact of dimension of rectangular aperture on shielding performance by using RBFNN and MLP network model with ANN has been studied, as a novelty.
Źródło:
International Journal of Electronics and Telecommunications; 2019, 65, 3; 359-365
2300-1933
Pojawia się w:
International Journal of Electronics and Telecommunications
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Head-Related Transfer Function Selection Using Neural Networks
Autorzy:
Yao, S.-N.
Collins, T.
Liang, C.
Powiązania:
https://bibliotekanauki.pl/articles/176307.pdf
Data publikacji:
2017
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Tematy:
head-related transfer function
neural networks
localisation
music
audio
anthropometry
pinna
Opis:
In binaural audio systems, for an optimal virtual acoustic space a set of head-related transfer functions (HRTFs) should be used that closely matches the listener’s ones. This study aims to select the most appropriate HRTF dataset from a large database for users without the need for extensive listening tests. Currently, there is no way to reliably reduce the number of datasets to a smaller, more manageable number without risking discarding potentially good matches. A neural network that estimates the appropriateness of HRTF datasets based on input vectors of anthropometric measurements is proposed. The shapes and sizes of listeners’ heads and pinnas were measured using digital photography; the measured anthropometric parameters form the feature vectors used by the neural network. A graphical user interface (GUI) was developed for participants to listen to music transformed using different HRTFs and to evaluate the fitness of each HRTF dataset. The listening scores recorded were the target outputs used to train the neural networks. The aim was to learn a mapping between anthropometric parameters and listener’s perception scores. Experimental validations were performed on 30 subjects. It is demonstrated that the proposed system produces a much more reliable HRTF selection than previously used methods.
Źródło:
Archives of Acoustics; 2017, 42, 3; 365-373
0137-5075
Pojawia się w:
Archives of Acoustics
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Neural networks for function approximation in dynamic modelling
Autorzy:
Nedbálek, J.
Powiązania:
https://bibliotekanauki.pl/articles/2069707.pdf
Data publikacji:
2008
Wydawca:
Uniwersytet Morski w Gdyni. Polskie Towarzystwo Bezpieczeństwa i Niezawodności
Tematy:
reliability
Monte Carlo
RBF neural network
simulation
temperature
Opis:
The paper demonstrates the comparsion of Monte Carlo simulation (MC) algorithm with the Radial Basis Function (RBF) neural network enhancement of the same algorithm in the reliability case study. In our test, we dispose of the tank containing liquid water – its temperature process variable evolves deterministicly according to the differential equation, which solution is known. All component failures are considered as a stochastic events. In the case of surpassing temperature treshhold of the liquid inside the tank, we interpret the situation as the system failure. With regard to process dynamics, we attempt to evaluate the tank system unreliability related to the initiative input parameters setting. The neural network is used in equation coeficients calculation, which is executed in each transient state. Due to the neural networks, for some of the initial component settings, we can achieve the results of computation faster than in classical way of coeficients calculating and substituting into the equation.
Źródło:
Journal of Polish Safety and Reliability Association; 2008, 2; 255--259
2084-5316
Pojawia się w:
Journal of Polish Safety and Reliability Association
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Lokalizacja punktów pomiarowych w systemie do trójwymiarowego pozycjonowania ciała wybranymi metodami sztucznej inteligencji
Detection of measurement points in a 3D body positioning system by means of artificial intelligence
Autorzy:
Czechowicz, A.
Tokarczyk, R.
Powiązania:
https://bibliotekanauki.pl/articles/131086.pdf
Data publikacji:
2009
Wydawca:
Stowarzyszenie Geodetów Polskich
Tematy:
fotogrametria
pozycjonowanie ciała
sieci neuronowe
perceptron wielowarstwowy
wsteczna propagacja błędów
sieci z radialnymi funkcjami bazowymi
photogrammetry
body positioning
neural networks
multi-layer perceptron
error back-propagation
radial basis function networks
Opis:
Fotogrametryczny system cyfrowy do pomiaru ciała ludzkiego dla celów badania wad postawy służy do wyznaczania przestrzennego położenia wybranych jego punktów. Wymaga on pomierzenia na zdjęciach cyfrowych trzech grup punktów, zwanych w tytule referatu punktami pomiarowymi: fotopunktów, markerów sygnalizowanych na pacjencie oraz źrenic oczu. Fotopunkty to czarno-białe sygnały pozwalające na orientację w przestrzeni modelu utworzonego ze zdjęć. Markery to styropianowe kulki o średnicy 4÷5 mm sygnalizujące wybrane elementy kośćca umieszczone na powierzchni ciała. Artykuł dotyczy wykorzystania sieci neuronowych do lokalizacji fotopunktów i styropianowych markerów. Zadaniem sieci jest klasyfikacja kolejnych fragmentów obrazu na zawierające obraz fotopunktu, markera lub niezawierające obrazu żadnego z nich. W ramach badań sprawdzono możliwość przeprowadzenia zdefiniowanej powyżej klasyfikacji sieciami o architekturze wielowarstwowego perceptronu (ang. Multi Layer Perceptron –MLP) ze wsteczną propagacją błędu oraz sieciami z radialnymi funkcjami bazowymi RBF (ang. Radial Basis Function Networks). Zweryfikowano przydatność reprezentacji opartej na informacji o rozkładzie wartości gradientu oraz jego kierunku dla celów wykrycia punktów pomiarowych. Wspomniana reprezentacja wywodzi się z badań nad selekcją podobrazów dla potrzeb dopasowania zdjęć lotniczych.
A digital photogrammetric system for making measurements of the human body for the purpose of studying faulty posture is designed to determine the three-dimensional location of selected points in the human body. It requires the measurement of three groups of points on digital images, points referred to in this paper’s title as measurement points, i.e. control points, markers indicated on the patient’s body and pupils of the eyes. Control points are black and white signals permitting the correct orientation in space of a model created from the images. The markers are balls of polystyrene foam of 4-5 mm diameter, placed on the body, which indicate selected elements of the human skeleton. This paper describes the utilisation of neural networks to locate control points and markers. The aim of the networks is to classify consecutive fragments of an image as containing control points, containing markers or not containing any of these features. The research covered evaluation of the possibility of conducting this classification using Multi Layer Perceptron Networks with back propagation of errors as well as with Radial Basis Function Networks. The usefulness of a representation based on information about the distribution of gradient value and direction for the purpose of the detection of measurement points has been verified. This representation comes from earlier research on the selection of subimages for the purpose of matching the aerial pictures.
Źródło:
Archiwum Fotogrametrii, Kartografii i Teledetekcji; 2009, 20; 67-79
2083-2214
2391-9477
Pojawia się w:
Archiwum Fotogrametrii, Kartografii i Teledetekcji
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Comparison of Second Order Algorithms for Function Approximation with Neural Networks
Autorzy:
Boutalbi, E.
Ait Gougam, L.
Mekideche-Chafa, F.
Powiązania:
https://bibliotekanauki.pl/articles/1402013.pdf
Data publikacji:
2015-08
Wydawca:
Polska Akademia Nauk. Instytut Fizyki PAN
Tematy:
02.30.Mv
07.05.Mh
Opis:
The Neural networks are massively parallel, distributed processing systems representing a new computational technology built on the analogy to the human information processing system. They are usually considered as naturally parallel computing models. The combination of wavelets with neural networks can hopefully remedy each other's weaknesses, resulting in wavelet based neural network capable of approximating any function with arbitrary precision. A wavelet based neural network is a nonlinear regression structure that represents nonlinear mappings as the superposition of dilated and translated versions of a function, which is found both in the space and frequency domains. The desired task is usually obtained by a learning procedure which consists in adjusting the "synaptic weights". For this purpose, many learning algorithms have been proposed to update these weights. The convergence for these learning algorithms is a crucial criterion for neural networks to be useful in different applications. In this paper, we use different training algorithms for feed forward wavelet networks used for function approximation. The training is based on the minimization of the least-square cost function. The minimization is performed by iterative first and second order gradient-based methods. We make use of the Levenberg-Marquardt algorithm to train the architecture of the chosen network and, then, the training procedure starts with a simple gradient method which is followed by a BFGS (Broyden, Fletcher, Glodfarb et Shanno) algorithm. The conjugate gradient method is then used. The performances of the different algorithms are then compared. It is found that the advantage of the last training algorithm, namely, conjugate gradient method, over many of the other optimization algorithms is its relative simplicity, efficiency and quick convergence.
Źródło:
Acta Physica Polonica A; 2015, 128, 2B; B-271-B-272
0587-4246
1898-794X
Pojawia się w:
Acta Physica Polonica A
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Sieci grzybowe - struktura, funkcje i wykorzystanie przez człowieka
Fungal networks - structure, function and use by humans
Autorzy:
Dominiak, Martyna
Lembicz, Marlena
Powiązania:
https://bibliotekanauki.pl/articles/1033858.pdf
Data publikacji:
2018
Wydawca:
Polskie Towarzystwo Przyrodników im. Kopernika
Tematy:
fungal network
long-distance translocation
mycorrhiza
sieci grzybowe
translokacja długodystansowa
mikoryza
Opis:
Grzyby to organizmy występujące we wszystkich strefach klimatycznych, zasiedlające głównie lądy. Dzięki dopasowującym się do warunków środowiska mechanizmom wzrostu, tworzą podziemne sieci, zajmujące znaczną powierzchnię. W obrębie sieci rosnącej w heterogenicznym środowisku zachodzi transport związków odżywczych przez translokację długodystansową. Translokacja ma kluczowe znaczenie dla przetrwania grzybni, ponieważ strzępki rosnące w rejonie ubogim w pokarm są wspierane przez znajdujące się w części zasobniejszej. Grzyby mogą wchodzić w interakcje z innymi organizmami. Wykorzystując czynniki Myc aktywują zespoły genów roślinnych, co umożliwia rozwój grzybni, kolonizację korzeni rośliny, a w efekcie prowadzi do zawiązania mikoryzy. Sieci mikoryzowe wykorzystywane są przez rośliny do komunikacji i ostrzegania się przed niebezpieczeństwem. Natomiast ludzie wykorzystują właściwości sieci grzybowych m.in. do planowania przebiegu sieci komunikacyjnych, mykoremediacji czy produkcji opakowań biodegradowalnych. Przyjmując, że na świecie występuje ok 1,5 miliona gatunków grzybów, z czego znanych jest jedynie ok. 10%, możemy przypuszczać, jak wiele ich niezwykłych właściwości pozostaje do odkrycia.
Fungi are mostly terrestrial organisms occurring in all climatic zones. Thanks to the growth mechanisms that are adaptable to environmental conditions, they form underground networks covering large areas. Within a network that grows in heterogenic environment, nutrients are allocated through a long-distance translocation. Translocation is of a key importance for mycelium survival, because hyphae growing in a nutrient-poor place are supported by hyphae from a nutrient-rich area. Fungi may also enter into interactions with other organisms. Using Myc factors, they activate plant gene complexes, which enables the development of mycelium and colonization of plant roots leading to the development of mycorrhiza. Mycorrhizal networks are used by plants to communicate and warn each other of a danger. In turn, humans use the characteristics of fungal networks, among others, to design the flow of communication systems, for myco-remediation and production of biodegradable packing materials. Assuming that about 1.5 mln of fungal species occur in the world, out of which only some 10% are known, we can only presume how many unusual properties of fungi remain still undiscovered.
Źródło:
Kosmos; 2018, 67, 2; 313-318
0023-4249
Pojawia się w:
Kosmos
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Comparative Application of Radial Basis Function and Multilayer Perceptron Neural Networks to Predict Traffic Noise Pollution in Tehran Roads
Autorzy:
Mansourkhaki, A.
Berangi, M.
Haghiri, M.
Powiązania:
https://bibliotekanauki.pl/articles/124655.pdf
Data publikacji:
2018
Wydawca:
Polskie Towarzystwo Inżynierii Ekologicznej
Tematy:
MLP
RBF
neural network
noise prediction
road traffic noise
Opis:
Noise pollution is a level of environmental noise which is considered as a disturbing and annoying phenomenon for human and wildlife. It is one of the environmental problems which has not been considered as harmful as the air and water pollution. Compared with other pollutants, the attempts to control noise pollution have largely been unsuccessful due to the inadequate knowledge of its effects on humans, as well as the lack of clear standards in previous years. However, with an increase of traveling vehicles, the adverse impact of increasing noise pollution on human health is progressively emerging. Hence, investigators all around the world are seeking to find new approaches for predicting, estimating and controlling this problem and various models have been proposed. Recently, developing learning algorithms such as neural network has led to novel solutions for this challenge. These algorithms provide intelligent performance based on the situations and input data, enabling to obtain the best result for predicting noise level. In this study, two types of neural networks – multilayer perceptron and radial basis function – were developed for predicting equivalent continuous sound level (LAeq) by measuring the traffic volume, average speed and percentage of heavy vehicles in some roads in west and northwest of Tehran. Then, their prediction results were compared based on the coefficient of determination (R2) and the Mean Squared Error (MSE). Although both networks are of high accuracy in prediction of noise level, multilayer perceptron neural network based on selected criteria had a better performance.
Źródło:
Journal of Ecological Engineering; 2018, 19, 1; 113-121
2299-8993
Pojawia się w:
Journal of Ecological Engineering
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Verification of applicability of the Trimble RTX satellite technology with xFill function in establishing surveying control networks
Weryfikacja przydatności technologii satelitarnej Trimble RTX z funkcją xFill do zakładania osnów pomiarowych
Autorzy:
Krzyżek, R.
Powiązania:
https://bibliotekanauki.pl/articles/145376.pdf
Data publikacji:
2013
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Tematy:
osnowa geodezyjna
technologia satelitarna
Trimble RTX
RTK GPS
geodetic control network
RTX
Opis:
The paper presents the results of real time measurements of test geodetic control network points using the RTK GPS and RTX Extended technologies. The Trimble RTX technology uses the xFill function, which enables real measurements without the need for constant connection with the ASG EUPOS system reference stations network. Comparative analyses of the results of measurements using the methods were performed and they were compared with the test control network data assumed to be error-free. Although the Trimble RTX technology is an innovative measurement method which is rarely used now, the possibilities it provides in surveying works, including building geodetic control networks, are satisfactory and it will certainly contribute to improving the organisation of surveying works.
W pracy przedstawiono wyniki pomiarów w czasie rzeczywistym punktów osnowy testowej z wykorzystaniem technologii RTK GPS oraz RTX Extended. W technologii Trimble RTX wykorzystano funkcję xFill, która daje możliwości realnego wykonywania pomiaru bez konieczności stałej łączności z siecią stacji referencyjnych systemu ASG EUPOS. Wykonano analizy porównawcze wyników pomiaru między metodami oraz odniesiono je do danych osnowy testowej, przyjętych za bezbłędne. Choć technologia Trimble RTX jest innowacyjną metodą pomiaru i jeszcze rzadko stosowaną, to możliwości jakie daje w realizacjach prac geodezyjnych, w tym zakładaniu osnów pomiarowych, są bardzo zadawalające i z pewnością przyczyni się do jeszcze lepszej i bardziej ekonomicznej organizacji prac geodezyjnych.
Źródło:
Geodesy and Cartography; 2013, 62, 2; 217-233
2080-6736
2300-2581
Pojawia się w:
Geodesy and Cartography
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Comparison of prototype selection algorithms used in construction of neural networks learned by SVD
Autorzy:
Jankowski, N.
Powiązania:
https://bibliotekanauki.pl/articles/330020.pdf
Data publikacji:
2018
Wydawca:
Uniwersytet Zielonogórski. Oficyna Wydawnicza
Tematy:
radial basis function network
extreme learning machine
kernel method
prototype selection
machine learning
k nearest neighbours
radialna funkcja bazowa
metoda jądrowa
uczenie maszynowe
metoda k najbliższych sąsiadów
Opis:
Radial basis function networks (RBFNs) or extreme learning machines (ELMs) can be seen as linear combinations of kernel functions (hidden neurons). Kernels can be constructed in random processes like in ELMs, or the positions of kernels can be initialized by a random subset of training vectors, or kernels can be constructed in a (sub-)learning process (sometimes by k-means, for example). We found that kernels constructed using prototype selection algorithms provide very accurate and stable solutions. What is more, prototype selection algorithms automatically choose not only the placement of prototypes, but also their number. Thanks to this advantage, it is no longer necessary to estimate the number of kernels with time-consuming multiple train-test procedures. The best results of learning can be obtained by pseudo-inverse learning with a singular value decomposition (SVD) algorithm. The article presents a comparison of several prototype selection algorithms co-working with singular value decomposition-based learning. The presented comparison clearly shows that the combination of prototype selection and SVD learning of a neural network is significantly better than a random selection of kernels for the RBFN or the ELM, the support vector machine or the kNN. Moreover, the presented learning scheme requires no parameters except for the width of the Gaussian kernel.
Źródło:
International Journal of Applied Mathematics and Computer Science; 2018, 28, 4; 719-733
1641-876X
2083-8492
Pojawia się w:
International Journal of Applied Mathematics and Computer Science
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Beta neuro-fuzzy systems
Autorzy:
Alimi, A. M.
Powiązania:
https://bibliotekanauki.pl/articles/1931568.pdf
Data publikacji:
2003
Wydawca:
Politechnika Gdańska
Tematy:
beta function
kernel based neural networks
Sugeno fuzzy model
neuro-fuzzy systems
universal approximation property
learning algorithms
incremental learning
Opis:
In this paper we present the Beta function and its main properties. A key feature of the Beta function, which is given by the central-limit theorem, is also given. We then introduce a new category of neural networks based on a new kernel: the Beta function. Next, we investigate the use of Beta fuzzy basis functions for the design of fuzzy logic systems. The functional equivalence between Beta-based function neural networks and Beta fuzzy logic systems is then shown with the introduction of Beta neuro-fuzzy systems. By using the SW theorem and expanding the output of the Beta neuro-fuzzy system into a series of Beta fuzzy-based functions, we prove that one can uniformly approximate any real continuous function on a compact set to any arbitrary accuracy. Finally, a learning algorithm of the Beta neuro-fuzzy system is described and illustrated with numerical examples.
Źródło:
TASK Quarterly. Scientific Bulletin of Academic Computer Centre in Gdansk; 2003, 7, 1; 23-41
1428-6394
Pojawia się w:
TASK Quarterly. Scientific Bulletin of Academic Computer Centre in Gdansk
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Approximation properties of some two-layer feedforward neural networks
Autorzy:
Nowak, M. A.
Powiązania:
https://bibliotekanauki.pl/articles/255577.pdf
Data publikacji:
2007
Wydawca:
Akademia Górniczo-Hutnicza im. Stanisława Staszica w Krakowie. Wydawnictwo AGH
Tematy:
neural networks
approximation of functions
sigmoidal function
Opis:
In this article, we present a multiyariate two-layer feedforward neural networks that approximate continuos functions defined on [0, 1]d. We show that the L1 error of approximation is asymptotically proportional to the modulus of continuity of the underlying function taken at √d/n, where n is the number of function values used.
Źródło:
Opuscula Mathematica; 2007, 27, 1; 59-72
1232-9274
2300-6919
Pojawia się w:
Opuscula Mathematica
Dostawca treści:
Biblioteka Nauki
Artykuł
Wyświetlanie 1-15 z 47

Ta witryna wykorzystuje pliki cookies do przechowywania informacji na Twoim komputerze. Pliki cookies stosujemy w celu świadczenia usług na najwyższym poziomie, w tym w sposób dostosowany do indywidualnych potrzeb. Korzystanie z witryny bez zmiany ustawień dotyczących cookies oznacza, że będą one zamieszczane w Twoim komputerze. W każdym momencie możesz dokonać zmiany ustawień dotyczących cookies