Design Of Radial Basis Function Networks Research Articles

Parameter control for differential evolution by storage of successful values at an individual level

Setting the parameters of differential evolution to appropriate values is critical for attaining satisfactory performance on a given problem (class). Therefore, parameter tuning is typically necessary, at least to some extent. However, even despite extensive tuning, the best possible performance may still not be achieved since different parameter values might be necessary during the different phases of the optimisation process. A solution to this issue is offered by parameter control schemes that properly adjust the parameters during the algorithm run. This paper presents a new self-adaptive parameter control scheme, which is based on the storage of a few previously successful values for each population member separately. These are then subsequently used to generate new and better-fitting parameter values with the aim of maintaining an effective search throughout the algorithm run. The highly competitive results obtained on challenging benchmark functions of different dimensionality suggest its viability and robustness. This is further corroborated by the promising results obtained on the problem of automatic radial basis function network design for classification needs.

Estimation of tyre forces using smart tyre sensors and artificial intelligence

In-tyre strain measurements from a smart tyre sensor system are analysed using two artificial neural network types, in order to estimate tyre forces. A tyre finite element model is used to calculate in-tyre strain (inputs) and tyre forces developed at the wheel centre (outputs) for use in the neural networks. Neural networks are trained on pure slip conditions and tested on combined slip events with the goal of accurately predicting tyre longitudinal and lateral forces and the aligning moment. The large mapping function is fitted using multilayer perceptron and radial basis function networks. Results from the best radial basis function network design are excellent, with training times under one minute, testing times of milli-seconds and calculated tyre forces within 1%. The conclusion is that radial basis function networks can be used effectively for real time analysis of strain sensor measurements in a smart tyre sensor system.

Estimation of tyre forces using smart tyre sensors and artificial intelligence

In-tyre strain measurements from a smart tyre sensor system are analysed using two artificial neural network types, in order to estimate tyre forces. A tyre finite element model is used to calculate in-tyre strain (inputs) and tyre forces developed at the wheel centre (outputs) for use in the neural networks. Neural networks are trained on pure slip conditions and tested on combined slip events with the goal of accurately predicting tyre longitudinal and lateral forces and the aligning moment. The large mapping function is fitted using multilayer perceptron and radial basis function networks. Results from the best radial basis function network design are excellent, with training times under one minute, testing times of milli-seconds and calculated tyre forces within 1%. The conclusion is that radial basis function networks can be used effectively for real time analysis of strain sensor measurements in a smart tyre sensor system.

Designing RBF Networks Using the Agent-Based Population Learning Algorithm

Radial Basis Function Neural Networks (RBFNs) are nowadays quite popular due to their ability to discover and approximate complex nonlinear dependencies within the data under analysis. Performance of the RBF network depends on numerous factors related to its initialization and training. The paper proposes an approach to the radial basis function networks design, where initial parameters of the network, output weights and parameters of the transfer function are set using the proposed agent-based population learning algorithm (PLA). The algorithm is validated experimentally. Advantages and main features of the PLA-based RBF designs are discussed basing on results of the computational experiment.

AGENT-BASED APPROACH TO THE DESIGN OF RBF NETWORKS

This article proposes a novel approach to the radial basis function network (RBFN) design. Its main idea is to apply the agent-based population learning algorithm to the task of initialization and training RBFNs. The approach allows for an effective network initialization and estimation of its output weights. The initialization involves two stages, where in the first one initial clusters are produced using the similarity-based procedure and next, in the second stage, prototypes (centroids) from the thus-obtained clusters are selected. The agent-based population learning algorithm is used to select prototypes. In the proposed implementation of the algorithm, both tasks—RBFN initialization and RBFN training—are carried out by a team of agents executing various local search procedures and cooperating with a view to determine the solution to the RBFN design problem at hand. The performance of the RBFN constructed using the proposed agent-based approach is analyzed and evaluated. The proposed approach is also compared with different RBFN initialization and training procedures in the literature.

PROTEIN SUPERFAMILY CLASSIFICATION USING ADAPTIVE EVOLUTIONARY RADIAL BASIS FUNCTION NETWORK

In this paper, the concept of adaptive multiobjective genetic algorithm (AMOGA) is applied for the structure optimization of radial basis function network (RBFN). The problem of finding the number of hidden centers remains a critical issue in the design of RBFN. The number of basis function controls the complexity and generalization ability of the network. The most parsimonious network obtained from the pareto front is applied in one of the challenging research area of proteomics and computational biology: Protein superfamily classification. The problem deals with predicting the family membership of a newly discovered amino acid sequence. The modification to the earlier approach of multiobjective genetic algorithm (MOGA) is done based on the two key controlling parameters such as probability of crossover and probability of mutation. These values are adaptively varied based on the performance of the algorithm i.e., based on the percentage of total population present in the best nondomination level. Principal component analysis (PCA) is used for dimension reduction and significant features are extracted from long feature vector of amino acid sequences. Numerical simulation results illustrates the efficiency of our approach in terms of faster convergence, optimal architecture and high level of classification accuracy.

RBF-Based QP Estimation Model for VBR Control in H.264/SVC

In this paper, we propose a novel variable bit rate (VBR) controller for real-time H.264/scalable video coding (SVC) applications. The proposed VBR controller relies on the fact that consecutive pictures within the same scene often exhibit similar degrees of complexity, and consequently should be encoded using similar quantization parameter (QP) values for the sake of quality consistency. In order to prevent unnecessary QP fluctuations, the proposed VBR controller allows for just an incremental variation of QP with respect to that of the previous picture, focusing on the design of an effective method for estimating this QP variation. The implementation in H.264/SVC requires to locate a rate controller at each dependency layer (spatial or coarse grain scalability). In particular, the QP increment estimation at each layer is computed by means of a radial basis function (RBF) network that is specially designed for this purpose. Furthermore, the RBF network design process was conceived to provide an effective solution for a wide range of practical real-time VBR applications for scalable video content delivery. In order to assess the proposed VBR controller, two real-time application scenarios were simulated: mobile live streaming and Internet protocol television broadcast. It was compared to constant QP encoding and a recently proposed constant bit rate (CBR) controller for H.264/SVC. The experimental results show that the proposed method achieves remarkably consistent quality, outperforming the reference CBR controller in the two scenarios for all the spatio-temporal resolutions considered.

A study on the medium-term forecasting using exogenous variable selection of the extra-virgin olive oil with soft computing methods

Time series forecasting is an important task for the business sector. Agents involved in the olive oil sector consider that, for the olive oil price, medium-term predictions are more important than short-term predictions. In collaboration with these agents the forecasting of the price of extra-virgin olive oil six months ahead has been established as the aim of this work. According to expert opinion, the use of exogenous variables and technical indicators can help in this task and must be included in the forecasting process. The amount of variables that can be considered makes necessary the use of feature selection algorithms in order to reduce the number of variables and to increase the interpretability and usefulness of the obtained forecasting system. Thus, in this paper CO2RBFN, a cooperative-competitive algorithm for Radial Basis Function Network design, and other soft computing methods have been applied to the data sets with the whole set of input variables and to the data sets with the selected set of input variables. The experimentation carried out shows that CO2RBFN obtains the best results in medium term forecasting for olive oil prices with the whole and with the selected set of input variables. Moreover, the feature selection methods applied to the data sets highlighted some influential variables which could be considered not only for the prediction but also for the description of the complex process involved in the medium-term forecasting of the olive oil price.

A DOE based approach for the design of RBF artificial neural networks applied to prediction of surface roughness in AISI 52100 hardened steel turning

The use of artificial neural networks for prediction in hard turning has received considerable attention in literature. An often quoted drawback of ANNs is the lack of a systematic way for the design of high performance networks. This study presents a DOE based approach for the design of ANNs of Radial Basis Function (RBF) architecture applied to surface roughness prediction in turning of AISI 52100 hardened steel. Experimental factors are the number of radial units on the hidden layer, the algorithm employed to calculate the spread factor of radial units and the algorithm employed to calculate radial function centers. DOE is employed to select levels of factors that benefit network prediction skills. Experiments with data sets of distinct sizes were conducted and network configurations leading to high performance were identified. ANN models obtained proved capable to predict roughness in accurate, precise and affordable way. Results pointed significant factors for network design and revealed that interaction effects between design parameters have significant influence on network performance for the task proposed. The work concludes that the DOE methodology constitutes a better approach to the design of RBF networks for roughness prediction than the most common trial and error approach.

Analysis of an evolutionary RBFN design algorithm, CO 2RBFN, for imbalanced data sets

In the classification problem field, we often encounter many real application areas in which the data do not have an equitable distribution among the different classes of the problem. In such cases, we are dealing with the so-called imbalanced data sets. This scenario has significant interest since standard classifiers are often biased towards the majority classes, whereas the minority ones tend to have a higher reward as they usually define the concepts of interest from the learning point of view. The aim of this paper is to analyse the performance of CO 2RBFN, a evolutionary cooperative–competitive model for the design of radial-basis function networks applied to classification problems on imbalanced domains, and to study its cooperation with a well-known pre-processing method, the “synthetic minority over-sampling technique”. The good performance of CO 2RBFN is shown through an experimental study carried out on a large collection of imbalanced data sets where we compare, by means of a proper statistical study, the behaviour of our model with many representative neural networks algorithms, the C4.5 decision tree and a hierarchical fuzzy rule-based classification system.

Design of Radial Basis Function Network Using Adaptive Particle Swarm Optimization and Orthogonal Least Squares

This paper presents a two-level learning method for designing an optimal Radial Basis Function Network (RBFN) using Adaptive Velocity Update Relaxation Particle Swarm Optimization algorithm (AVURPSO) and Orthogonal Least Squares algorithm (OLS) called as OLS-AVURPSO method. The novelty is to develop an AVURPSO algorithm to form the hybrid OLS-AVURPSO method for designing an optimal RBFN. The proposed method at the upper level finds the global optimum of the spread factor parameter using AVURPSO while at the lower level automatically constructs the RBFN using OLS algorithm. Simulation results confirm that the RBFN is superior to Multilayered Perceptron Network (MLPN) in terms of network size and computing time. To demonstrate the effectiveness of proposed OLS-AVURPSO in the design of RBFN, the Mackey-Glass Chaotic Time-Series as an example is modeled by both MLPN and RBFN.

CO2RBFN: an evolutionary cooperative–competitive RBFN design algorithm for classification problems

This paper presents a new evolutionary cooperative---competitive algorithm for the design of Radial Basis Function Networks (RBFNs) for classification problems. The algorithm, CO2RBFN, promotes a cooperative---competitive environment where each individual represents a radial basis function (RBF) and the entire population is responsible for the final solution. The proposal considers, in order to measure the credit assignment of an individual, three factors: contribution to the output of the complete RBFN, local error and overlapping. In addition, to decide the operators' application probability over an RBF, the algorithm uses a Fuzzy Rule Based System. It must be highlighted that the evolutionary algorithm considers a distance measure which deals, without loss of information, with differences between nominal features which are very usual in classification problems. The precision and complexity of the network obtained by the algorithm are compared with those obtained by different soft computing methods through statistical tests. This study shows that CO2RBFN obtains RBFNs with an appropriate balance between accuracy and simplicity, outperforming the other methods considered.

Visual RBF network design based on Star Coordinates

The performance of neural networks is greatly affected by their design, yet the question of finding optimal designs remains open and inspires a considerable amount of research. Most of the researches have been focused on developing automatic algorithms for neural network configuration. This paper addresses the problem of Radial Basis Function Network (RBFN) design definition with a visual technique, called Star Coordinates. The purpose of this approach is to enable the RBFN design revision and refining process, capitalizing on the power of visualization and interactive operations.

Application of radial basis function network in variable blank holding force identification

The strategy of variable blank holding force is an effective measure for the springback control in sheet metal forming processes.The Radial Basis Function(RBF)network and three-dimensional springback simulating technology were employed for variable blank holding force from the forming effects.Meanwhile,the research on the dynamic structure design of RBF network was done and the dynamic structure design method,Dynamic Structure Design of Radial Basis Function network(DYNSDRBF),based on generalization performance was developed and programmed.The application results of DYNSDRBF in the variable blank holding force identification indicate that the structure design method can show better calculation accuracy.The research results on variable blank holding force identification show that the solving method is helpful for improving the sheet forming effects.

Automatical initialization of RBF neural networks

Although many methods are devoted to the design of Radial Basis Function Networks (RBFN), the lack of automatic approaches makes it difficult to generate suitable models in industrial applications. The object of this paper therefore proposes a deterministic method able to automatically select leaders or prototypes on which the RBFN design can be developed. This technique combines clustering and fuzzy C-means algorithms adapted to supervised contexts, and was tested successfully in a real application for Medicinal Chemistry, which was a data set regrouping 581 molecules active in the Central Nervous System. A comparison between the results obtained by this approach and by other standard initialization methods showed that our algorithm clearly improved the classification ability of RBFN.