• Proceedings of the IEEK Conference
• /
• 2002.07a
• /
• pp.30-33
• /
• 2002

To obtain good performance of radial basis function (RBF) neural networks, it needs very careful consideration in design. The selection of several parameters such as the number of centers and widths of the radial basis functions must be considered carefully since they critically affect the network's performance. We propose a learning algorithm for growing of complexity of RBF neural networks which is adapted automatically according to the complexity of tasks. The algorithm generates a new basis function based on the errors of network, the percentage of decreasing rate of errors and the nearest distance from input data to the center of hidden unit. The RBF's center is located at the point where the maximum of absolute interference error occurs in the input space. The width is calculated based on the standard deviation of distance between the center and inputs data. The steepest descent method is also applied for adjusting the weights, centers, and widths. To demonstrate the performance of the proposed algorithm, general problem of function estimation is evaluated. The results obtained from the simulation show that the proposed algorithm for RBF neural networks yields good performance in terms of convergence and accuracy compared with those obtained by conventional multilayer feedforward networks.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.6
• /
• pp.1230-1237
• /
• 2009

In this paper, we develop a design methodology of Granular-based Radial Basis Function Neural Networks(GRBFNN) by context-based clustering. In contrast with the plethora of existing approaches, here we promote a development strategy in which a topology of the network is predominantly based upon a collection of information granules formed on a basis of available experimental data. The output space is granulated making use of the K-Means clustering while the input space is clustered with the aid of a so-called context-based fuzzy clustering. The number of information granules produced for each context is adjusted so that we satisfy a certain reconstructability criterion that helps us minimize an error between the original data and the ones resulting from their reconstruction involving prototypes of the clusters and the corresponding membership values. In contrast to "standard" Radial Basis Function neural networks, the output neuron of the network exhibits a certain functional nature as its connections are realized as local linear whose location is determined by the values of the context and the prototypes in the input space. The other parameters of these local functions are subject to further parametric optimization. Numeric examples involve some low dimensional synthetic data and selected data coming from the Machine Learning repository.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.5C
• /
• pp.452-460
• /
• 2003

In this paper, an equalizer(RNE) using nonlinear multilayer combining techniques in Bayesian equalizer with a structure of radial basis function network is proposed in order to simplify the structure and enhance the performance of the equalizer(RE) using a radial basis function network. The conventional RE Produces its output using linear combining the outputs of the basis functions in the hidden layer while the proposed RNE produces its output using nonlinear combining the outputs of the basis function in the first hidden layer. The nonlinear combiner is implemented by multilayer perceptrons(MLPs). In addition, as an infinite impulse response structure, the RNE with decision feedback equalizer (RNDFE) is proposed. The proposed equalizer has simpler structure and shows better performance than the conventional RE in terms of bit error probability and mean square error.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.6 no.2
• /
• pp.26-35
• /
• 1996

In this paper, some fundamental problems concerning RBF(radial-basis-function) networks and approximation of functions are addressed. First, a comprehensive introduction to RBF networks is given with typical RBF networks classified into three classes. Next, sharp conditions are given under which continuous functions of a finite number of real variables can be approximated arbitrarily well by a certain class of RBF networks. Finally, a related result is given concerning the representation of functions in the form of distributed RBF networks.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.26.6-26
• /
• 2001

In this paper we introduce a new algorithm that enables radial basis function network(RBFN) to be structured automatically and guarantees the stability of the RBFN. Because this new algorithm is efficient and also have the advantage of fast computational speed we adopt this algorithm as online learning scheme for uncertain nonlinear dynamical systems. Based on the fact that a 3-layered RBFN can represent a specific nonlinear function reasonably well by linearly combining a set of nonlinear and localized basis functions, we show that this RBFN can identify the nonlinear system very well without knowing the information of the system in advance.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.38 no.5
• /
• pp.1-6
• /
• 2001

In this paper, we propose the initial optimized structure of the Radial Basis Function Network(RBFN) which is more simple in the part of the structure and converges more faster than Neural Network. For this, we use the analysis method using time frequency localization and we can decide the initial structure of the RBFN suitable for the given problem. When we compose the hidden nodes with the radial basis functions whose localization are similar with the target function in the plane of the time and frequency, we can make a good decision of the initial structure having an ability of approximation.

• Transactions on Control, Automation and Systems Engineering
• /
• v.2 no.4
• /
• pp.268-273
• /
• 2000

Plasma models are crucial to equipment design and process optimization. A radial basis function network(RBFN) in con-junction with statistical experimental design has been used to model a process plasma. A 2$^4$ full factorial experiment was employed to characterized a hemispherical inductively coupled plasma(HICP) in characterizing HICP, the factors that were varied in the design include source power, pressure, position of shuck holder, and Cl$_2$ flow rate. Using a Langmuir probe, plasma attributes were collected, which include typical electron density, electron temperature. and plasma potential as well as their spatial uniformity. Root mean-squared prediction errors of RBEN are 0.409(10(sup)12/㎤), 0.277(eV), and 0.699(V), for electron density, electron temperature, and Plasma potential, respectively. For spatial uniformity data, they are 2.623(10(sup)12/㎤), 5.704(eV) and 3.481(V), for electron density, electron temperature, and plasma potential, respectively. Comparisons with generalized regression neural network(GRNN) revealed an improved prediction accuracy of RBFN as well as a comparable performance between GRNN and statistical response surface model. Both RBEN and GRNN, however, experienced difficulties in generalizing training data with smaller standard deviation.

• Journal of KIISE:Software and Applications
• /
• v.31 no.5
• /
• pp.555-562
• /
• 2004

Software industry has put more emphasis on maintenance and enhancement work than on the new development. The existing effort estimation models can still be applied to maintenance projects, though modifications are needed. This paper suggests a way to estimate the size of a maintenance project from the regression analysis of ISBSG's benchmarking data. First of all, among the 3 elements(addition, modification and deletion of the program) which influences the software cost, we selected and classified 4 groups from a total of 8 which shows actual maintenance cost from ISBSG's data. Moreover, we developed statistical model and a model which uses RBF(Radial Basis Function) Network and after evaluating each functions we concluded that the RBF Network is superior to the statistical model.

• Geomechanics and Engineering
• /
• v.33 no.6
• /
• pp.567-581
• /
• 2023

The prediction and achievement of a proper rock fragmentation size is the main challenge of blasting operations in surface mines. This is because an optimum size distribution can optimize the overall mine/plant economics. To this end, this study attempts to develop four improved artificial intelligence models to predict rock fragmentation through cascaded forward neural network (CFNN) and radial basis function neural network (RBFNN) models. In this regards, the CFNN was trained by the Levenberg-Marquardt algorithm (LMA) and Conjugate gradient backpropagation (CGP). Further, the RBFNN was optimized by the Dragonfly Algorithm (DA) and teaching-learning-based optimization (TLBO). For developing the models, the database required was collected from the Midouk copper mine, Iran. After modeling, the statistical functions were computed to check the accuracy of the models, and the root mean square errors (RMSEs) of CFNN-LMA, CFNN-CGP, RBFNN-DA, and RBFNN-TLBO were obtained as 1.0656, 1.9698, 2.2235, and 1.6216, respectively. Accordingly, CFNN-LMA, with the lowest RMSE, was determined as the model with the best prediction results among the four examined in this study.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.37 no.5
• /
• pp.433-441
• /
• 2009

Generally, the events in nature have multi-disciplinary characteristics. To solve this problems, these days loosely coupled methods are widely applied because of advantage of solvers which are already developed and well proved. Those solvers use different mesh system, so transformation and mapping of data are vital in the field of fluid-structure interaction(FSI). In this paper, the interpolation of deformation which is used globally and compactly supported radial basis functions(RBF), and mapping of force which use principle of virtual work are examined for computing time and accuracy to compare ability with simple 3-D problem. As the results, interpolation scheme of compactly supported radial basis functions are useful to interpolation and mapping for large-scale airplane in FSI with a k-dimensional tree(kd-tree) which is a space-partitioning data structure for organizing points in a k-dimensional space.