Search

In this study, a novel approach, interval fuzzy c-regression models with competitive agglomeration (IFCRMCA), is proposed to deal with the symbolic interval-valued data. The proposed IFCRMCA approach can identify the partition of the interval-valued data using both the distances to the cluster centers and the errors of interval regression models for each cluster. Due to the concepts of competitive agglomeration is used in the proposed approach, the pre-determination of the cluster number in the proposed IFCRMCA is not necessary. Various real experiments are carried on and the experimentally results shows that the proposed approaches are superior to the existing approaches.

In this paper, a novel interval possibilistic fuzzy c-means (IPFCM) clustering method is proposed for clustering symbolic interval data. Clustering algorithms are important methods that required in pattern recognition, data mining and text mining, etc. Most of cluster algorithms focus on single-valued data. The interval fuzzy c-means (IFCM) clustering method is first proposed for symbolic interval data. However, the IFCM may not have good performance when facing noisy date or data with outliers. Hence, the advantages of the proposed IPFCM clustering algorithm in this study mainly overcome the noisy date or data with outliers in symbolic interval data. In the proposed approach, two different distances; namely, Euclidean and Hausdorff distance measures are all individually considered. That is, based on Lagrange multipliers nonlinear programming method, the proposed IPFCM clustering algorithm under Euclidean and Hausdorff distance measure was individually derived for symbolic interval data. From our experimental results, the proposed IPFCM clustering algorithm indeed has better performance than the IFCM and the interval fuzzy possibilistic c-means (IFPCM) clustering algorithm under different distance measures for the noisy date or data with outlier problem. Besides, the proposed IPFCM clustering algorithm is also implemented on windows smart phone platform and shown better performance as expected. Finally, a real data set is added for testing.

In order to improve associative memory in artificial brain for the fuzzy photo and amnesia, we apply nonlinear principal component analysis (NLPCA) based deep learning to improve the problem of associative memory on capability. That is, the proposed NLPCA based deep associative memory neural networks can improve auto and bidirectional associative memory on capability problem in this paper. We proposed the seven-layered multi-layer perceptron neural network based on NLPCA and used a TensorFlow module with python to implement the proposed method. The proposed method used back-propagation to replace Hebbian learning for bidirectional associative memory. Finally, we presented seven-layered multi-layer perceptron neural networks under NLPCA with 0~9 numbers in patterns for the proposed method.

CBR (Case-based reasoning) is an effective reasoning mechanism that solves a new problem by remembering a previous similar knowledge and by reusing information and knowledge of that features. The case-base is a set of small case-bases. Each small case-base can be viewed as the result of granular computing. To obtain an efficient CBR system, this paper proposed a hybrid CBR system by introducing feature selection and Granular computing, it also incorporate similarity margin concept and Gaussian kernel fuzzy rough sets in case-based organization.

In this paper, a novel approach is introduced for tracking human targets in cases of high influence from complexity of environment. In real world applications, the need of interaction in real time between a tracking system and human plays an important role. In the proposed approach, a RGB-D camera is utilized to acquire the depth information which is considered to define the Depth Of Interest (DOI). This DOI is used to combine with the CAM-shift algorithm in human tracking. The Kalman filter is also implemented to help in predicting the direction of target. Comparing with the original CAM-shift algorithm, our approach performs more accurate and more effective results. The experiment results also show that our system can be implemented for real-time applications as well.

Less offspring and aging are global issues that will be a bad news for cleaning work that requires a lot of manpower; the nightly cleaning jobs for large-scale indoor spaces introduce workers' overwork/fatigue problem. This study proposes a robot suitable for night time cleaning, which uses the industrial computer as the software control core, the AVR MCU for the control hardware part, the various obstacles avoidance sensors, and the cleaning mode algorithms. In the section of mechanism design, with a rotating space detection capability, the robot can safely avoid the steering into the corner. Users can utilize the interface for remote control and remote monitoring, allowing users to see the details of the cleaning and handling of goods transportation; remote monitoring can provide the video of robot's cleaning, so that users knows the cleaning status. Large-scale space cleaning robots used in the late night hours manage to transfer the original late-night cleaning manpower to the peak hours, not only solve the problem of late-night work but also solve the problem of lack of manpower during peak hours.

Microarray data analysis approach has became a widely used tool for disease detection. It uses tens of thousands of genes as input dimension that would be a huge computational problem for data analysis. In this chapter, the proposed approach deals with selection of feature genes and classification of microarray data under support vector machine (SVM) approach. Feature genes can be finding out according to the adjustable epsilon-support vector regression (epsilon-SVR) and then to select high ranked genes after all microarray data. Moreover, multi-class support vector classification (multi-class SVC) and cross-validation methods apply to acquire great prediction classification accuracy and less computing time.

This paper demonstrates that radial basis function networks (RBFNs) with support vector regression (SVR) and annealing robust learning algorithm (ARLA) can be used effectively for the identification of the nonlinear dynamic systems with outliers. When the RBFNs are used for the identification of the nonlinear dynamic system, the number of hidden nodes, the initial parameters of the kernel, and the initial weights of the network must be determined first, a SVR approach is proposed to solve the initial problem of RBFNs. That is, the SVR uses the quadratic programming optimization to determine the initial structure of the RBFNs. Besides, the new cost function for the system identification with outliers is also proposed. That is, the proposed annealing robust radial basis function networks (ARRBFNs) are trained by the ARLA, which uses the annealing concept in the cost function of the robust back-propagation learning algorithm, can overcome the error measurement caused by the outliers. Simulation results show the superiority of the proposed method with different SVR.

This paper demonstrates an approach to predict the chaotic time series with outliers using annealing robust fuzzy neural networks (ARFNNs). A combination model that merges support vector regression (SVR), radial basis function networks (RBFNs) and simplified fuzzy inference system is used. The SVR has the good performances to determine the number of rules in the simplified fuzzy inference system and initial weights for the fuzzy neural networks (FNNs). Based on these initial structures, and then annealing robust learning algorithm (ARLA) can be used effectively to overcome outliers and adjust the parameters of structures. Simulation results show the superiority of the proposed method with different SVR for training and prediction of chaotic time series with outliers.

In this article, annealing robust radial basis function networks (ARRBFNs), which consist of a radial basis function network and a support vector regression (SVR), and an annealing robust learning algorithm (ARLA) are proposed for the prediction of chaotic time series with outliers. In order to overcome the initial structural problems of the proposed neural networks, the SVR is utilized to determine the number of hidden nodes, the initial parameters of the kernel, and the initial weights for the proposed ARRBFNs. Then the ARLA that can conquer the outliers is applied to tune the parameters of the kernel and the weights in the proposed ARRBFNs under the initial structure with SVR. The simulation results of Mackey-Glass time series show that the proposed approach with different SVRs can cope with outliers and give a fast learning speed. The results of the simulation are also given to demonstrate the validity of proposed method for chaotic time series with outliers.

In this study, CPBUM neural networks with annealing robust learning algorithm (ARLA) are proposed to improve the problems of conventional neural networks for modeling with outliers and noise. In general, the obtained training data in the real applications maybe contain the outliers and noise. Although the CPBUM neural networks have fast convergent speed, these are difficult to deal with outliers and noise. Hence, the robust property must be enhanced for the CPBUM neural networks. Additionally, the ARLA can be overcome the problems of initialization and cut-off points in the traditional robust learning algorithm and deal with the model with outliers and noise. In this study, the ARLA is used as the learning algorithm to adjust the weights of the CPBUM neural networks. It tunes out that the CPBUM neural networks with the ARLA have fast convergent speed and robust against outliers and noise than the conventional neural networks with robust mechanism. Simulation results are provided to show the validity and applicability of the proposed neural networks.

In this paper, to integrate android systems, intelligent computing and web service on an intelligent mobile web service computing systems for bioinformatics is proposed. That is, users can send parameters to control the intelligent computing system under Matlab/C#/Java and query from clientside to server-side in the proposed bioinformatics systems with their android phone. Besides, the proposed system can also deal with disease genes search under web service with bioinformatics web. At the same time, web service is important communication protocol in the proposed systems. Consequently, the proposed systems can provide that user can modify parameters for controlling the proposed intelligent computing, deal with disease genes search under web service with bioinformatics web and obtain results from the proposed systems by android phone arbitrarily.

Microarray data analysis approach has been became a widely used tool for disease detection. It uses tens of thousands of genes as input dimension that would be a huge computational problem for data analysis. In this paper, we proposed to apply affinity propagation (AP) clustering for feature genes selection of adult acute Lymphoblastic Leukemias (ALL) microarray data. That is, feature genes can be finding out according to the adjustable the number of cluster in AP clustering. AP Clustering is a new grouping method by passing messages between data points, AP clustering can to reduce dimension on each sample without known the number of cluster in advance. Finally, these results under the specific genes with AP clustering can provide learning in classification and prediction.

In recently, there are many machine learning approaches have developed for intelligent control. One of these approaches is least squares-support vector machine regression (LS-SVMR). Besides, the robustness problem of the LS-SVMR among machine learning algorithms is importantly considered in recent years. Hence, for the robustness problem in LS-SVMR, a least trimmed squares support vector machine regression (LTS-SVMR) with wavelet kernel which is the hybrid of the LTS and LS-SVMR is proposed in this paper. When the LTS method faces on the training sample with noise and outliers, it can effectively remove large noise and outliers under the proper initial nonlinear function. Hence, robustness problem in LS-SVMR is enhanced by combining the LS-SVMR with wavelet kernel and the LTS. Finally, the proposed LTS-SVMR with wavelet kernel is applied on modeling of nonlinear systems with noise and outliers.

In general, there are many machine learning algorithms have developed for the intelligent control. Besides, support vector machine (SVM) among machine learning is generally used in recent years. That is, many contributions about the support vector machine regression (SVMR) and the least squares-support vector machine regression (LS-SVMR) can be found in some well-known journals. In this paper, for the robustness problem of LS-SVMR, we propose the least trimmed squares support vector machine regression (LTS-SVMR) which is the hybrid of the least trimmed squares (LTS). That is, when the LTS method faces on the training sample with noise and outliers, it can effectively remove large noise and outliers under the proper initial nonlinear function. That is, robustness of the LS-SVMR is enhanced by combining the LS-SVMR and the LTS. Finally, the proposed LTS-SVMR is applied on modeling of nonlinear systems with noise and outliers.

Based on particle swarm optimization (PSO) algorithm for optimal design of PID controller, a method is proposed to navigate of a two-wheeled mobile robot from a given initial configuration to a desired final configuration in an unknown environment filled with obstacles. The mobile robot can follow a wall with PID control to adjust motor speed and avoid obstacles with moving square method in an unknown environment. The PSO algorithm is an optimization technique which is inspired by the social-like behavior, such as bird flocking or fish utilized it to solve optimization problems. It finds the global best solution by simply adjusting the trajectory of each particle its own best particle and toward the best particle of the entire swarm at each generation. The PSO algorithm becomes very popular due to it can be easily implemented and quickly find a good solution. Then the PSO algorithm can be used to adjust parameters of PID to control wheel velocities in an unknown environment filled with obstacles. To show the feasibility of the proposed method, simulation results and experiments are included for illustration.

A Chebyshev polynomial-based unified model (CPBUM) neural network is introduced and applied to control a magnetic bearing systems. First, we show that the CPBUM neural network not only has the same capability of universal approximator, but also has faster learning speed than conventional feedforward/recurrent neural network. It turns out that the CPBUM neural network is more suitable in the design of controller than the conventional feedforward/recurrent neural network. Second, we propose the inverse system method, based on the CPBUM neural networks, to control a magnetic bearing system. The proposed controller has two structures; namely, off-line and on-line learning structures. We derive a new learning algorithm for each proposed structure. The experimental results show that the proposed neural network architecture provides a greater flexibility and better performance in controlling magnetic bearing systems.

Recently, APPs (applications for intelligent mobile devices) about car recorder systems have been proposed for various mobile OS. Due to the numbers of APPs in the iOS is more than that of others OS, our car video recorder system is built on iOS. In addition, the objective-c programming language is employed to code APPs in the Xcode. Usually, the iOS mobile devices (e.g. iPhone) are equipped with GPS and camera. Hence, we only need to design software on the integration and links so as to replace the traditional tachograph. Basically, the intelligent mobile device has three types of positioning methods: built-in GPS, telecommunication base-station triangulation positioning method and Wi-Fi base-station positioning method. Although the user's position can be quickly located, it can not provide an accurately location for user when the positioning signal are hardly received. Besides, the speed data provided by the positioning methods may easily be affected by disturbances. It should be noted that iPhone usually cannot provide an accurate car's speed. In this study, a corrected speed method for the intelligent video car recorder is proposed. In the corrected speed method, the car speed in iPhone is corrected by the linear regression method. In our experience, the proposed method can improve the accuracy of the car speed estimation in iPhone.

This study is a numerical analysis with the peak power of sound channel under its numerical range of the region as the signal of voice communication can automatically start Bluetooth on iOS smart phone systems. Besides, the voice communication link using Bluetooth protocol. At the same time, the entire system is implemented in the intelligent mobile devices. In this study, the intelligent mobile devices use microphone and mobile device operating system's built-in audio API to collect the data from the peak power channel of voice. The collected data is trended and noise reduction processing, so that it is easy to be used to identify trends in data. From the data analysis of the experimental results show that the significant differences in peak power generated by the people blowing sound and spoken voice data. Then, it can use the peak power value as a voice call to control the start signal of Bluetooth. The user can automatically open the call process by the people blowing sound without any manually processes. Besides, the peak power identification model is also a universal model, it can be easily extended into other operating systems on smart phone.

In this study, build interactive location database intelligent video car recorder system. Video automotive logger system was established in the iOS intelligent mobile devices (iPhone). The video car recorder system can be record the current driving conditions, display the current location on the map, and upload it to the location database. Location database is constructed by a network server, MySQL, the position database server to create and link through PHP server-side language. In this study, the proposed system is constructed by the iOS SDK, google maps APIs, PHP and MySQL tools. Finally, the proposed system has been uploaded into Apple store.

In this paper, based on least trimmed squares-support vector regression (LTS-SVR), a robust radial basis function network (RRBFN) is proposed in the modeling problem to deal with training data sets that may contain outliers and noises. There are two stages in the proposed RRBFN approach. In stage I, outliers and large noises will be trimmed via the LTS-SVR procedure so that the influences of outliers and large noises can be reduced. In other words, with the use of the LTS-SVR procedure, the proposed approach can obtain an appropriate initial structure for the RRBFN to avoid possible overfitting phenomena. It can also be found that with this procedure it may result in a fast convergence speed. After stage I, the remaining data of the training data set are directly used to adjust the parameters of the RRBFN through the gradient-descent kind of learning algorithm. Hence, it does not need to take extra time to compute the weights of the robust cost function like that considered in the M-estimate kind of approaches. From our simulation results, the performances of the proposed system are superior to that of using the annealing robust RBFN and that of using the Wilcoxon generalized RBFN when training data contain outliers and noises.

In this paper, least trimmed squares (LTS) based CPBUM neural networks are proposed to improve the outliers and noise problems of conventional neural networks. In general, the obtained training data in the real applications maybe contain the outliers and noise. Although the CPBUM neural networks have fast convergent speed, this model is difficult to deal with training data set with outliers and noise. Hence, the robust property must be enhanced for the CPBUM neural networks. In this paper, the LTS computational architecture is proposed for the CPBUM neural networks. That is, the LTS approach can trim some large noise and outliers in the training data set. After the LTS, the gradient-descent kind of learning algorithms is used as the learning algorithm to adjust the weights of the CPBUM neural networks. It tunes out that the LTS based CPBUM neural networks have fast convergent speed and robust against outliers and noise than the conventional neural networks with robust mechanism. Simulation results are provided to show the validity and applicability of the proposed neural networks.

Modern information fusion systems essentially associate decision-making processes with multi-sensor systems. Precise decision-making processes depend upon aggregating useful information extracted from large numbers of messages or large datasets; meanwhile, the distributed multi-sensor systems which employ several geographically separated local sensors are required to provide sufficient messages or data with similar and/or dissimilar characteristics. These kinds of information fusion techniques have been widely investigated and used for implementing several information retrieval systems. However, the results obtained from the information fusion systems vary in different situations and performing intelligent aggregation and fusion of information from a distributed multi-source, multi-sensor network is essentially an optimization problem. A flexible and versatile framework which is able to solve complex global optimization problems is a valuable alternative to traditional information fusion. Furthermore, because of the highly dynamic and volatile nature of the information flow, a swift soft computing technique is imperative to satisfy the demands and challenges. In this paper, a nonlinear aggregation based on the Choquet integral (NACI) model is considered for information fusion systems that include outliers under inherent interaction among feature attributes. The estimation of interaction coefficients for the proposed model is also performed via a modified algorithm based on particle swarm optimization with quantum-behavior (QPSO) and the high breakdown value estimator, least trimmed squares (LTS). From simulation results, the proposed MQPSO algorithm with LTS (named LTS-MQPSO) readily corrects the deviations caused by outliers and swiftly achieves convergence in estimating the parameters of the proposed NACI model for the information fusion systems with outliers.

Self-constructing neural fuzzy inference network (SONFIN) is a neural fuzzy system and owing to its structure learning capability, SONFIN has been demonstrated to have excellent learning performance. However, various parameters must be selected in the implementation of SONFIN. In this paper, the learning behavior of SONFIN is studied. First, the SONFIN system with different thresholds and variances are considered. Different selections will result in different rule numbers and membership function widths. Our experiment results indicate that when there are possibilities of overfitting, more rules may not always come up with better performance. Secondly, two different learning algorithms are considered; the backpropagation (BP) learning algorithm and the recursive least square (RLS) algorithm. It can found that the learning of using RLS is much faster than that of using BP as expected. However, it can be found that when overfitting may occur, BP can have better learning performance in terms of testing errors. Finally, the use of reset for the covariance matrix in the RLS algorithm is investigated. From this primitive study, it can be found that the learning algorithms or some parameter selections may have both good effects in the testing performance and the training performance before the learning does not have significant overfitting. However, when the learning crosses this point, any selection is good for learning may bring bad effects on the testing performance.

The research on theory and application of magnetic bearing system has received much attention in recent years. In order to overcome nonsystematic design and long training time in nonlinear adaptive inverse control, we propose a new method based on CPBUM neural network. The experimental results show that the proposed architecture provides a greater flexibility and better performance in controlling magnetic bearing system.

In this paper, the hybrid support vector machines (SVM) and Gaussian process (GPs) are proposed to deal with the molecular autoregulatory feedback loop systems with outliers. In the proposed approach, there are two-stage strategies. In the stage 1, the support vector machine regression (SVMR) approach is used to filter out the outliers in the training data set. Because of the large outliers in the training data set are almost removed, the large outlier's effects are reduce, so the concepts of robust statistic theory are not used to reduce the outlier's effects. The rest of the training data set after the stage 1 is directly used to training the Gaussian process for regression (GPR) in the stage 2. According to the simulation results, the performance of the proposed approach is superior to the least squares support vector machines for regression, and GPR when the outliers are existed in the molecular autoregulatory feedback loop systems.

In this study, the robust least square support vector machines for regression (RLS-SVMR) is proposed to deal with training data set with outliers. There are two-stage strategies in the proposed approach. In the stage I, called as data preprocessing, the support vector regression (SVR) approach is used to filter out the outliers in the training data set. Due to the outliers in the training data set are removed, the concepts of robust statistic theory have no need to reduce the outlierpsilas effect. Then, the training data set except for outliers, called as the reduced training data set, is directly used to training the non-robust least squares support vector machines for regression (LS-SVMR) in the stage II. Consequently, the learning mechanism of the proposed approach is much easier than the weighted LS-SVMR approach. Based on the simulation results, the performance of the proposed approach is superior to the weighted LS-SVMR approach when the outliers are existed.

In this study, the hybrid support vector machines for regression (HSVMR) is proposed to deal with training data set with outliers for fuzzy neural networks (FNNs). There are two-stage strategies in the proposed approach. In the stage I, called as data preprocessing, the support vector machines for regression (SVMR) approach is used to filter out the outliers in the training data set. Due to the outliers in the training data set are removed, the concept of robust statistic theory have no need to reduce the outlierpsilas effect. Then, the training data set except for outliers, called as the reduced training data set, is directly used to training the sparse least squares support vector machines for regression (LS-SVMR) in the stage H. Consequently, the learning mechanism of the proposed approach for fuzzy neural network does not need iterated learning for simplified fuzzy inference systems. Based on the simulation results, the performance of the proposed approach is superior to the robust LS-SVMR approach when the outliers are existed.

Support vector regression (SVR) employs the support vector machine (SVM) to tackle problems of function approximation and regression estimation. SVR has been shown to have good robust properties against noise. When the parameters used in SVR are improperly selected, overfitting phenomena may still occur. However, the selection of various parameters is not straightforward. Besides, in SVR, outliers may also possibly be taken as support vectors. Such an inclusion of outliers in support vectors may lead to seriously overfitting phenomena. In this paper, a novel regression approach, termed as the robust support vector regression (RSVR) network, is proposed to enhance the robust capability of SVR. In the approach, traditional robust learning approaches are employed to improve the learning performance for any selected parameters. From the simulation results, our RSVR can always improve the performance of the learned systems for all cases. Besides, it can be found that even the training lasted for a long period, the testing errors would not go up. In other words, the overfitting phenomenon is indeed suppressed.

Bioinformatics is the computing response to the molecular revolution in biology. This revolution has reshaped the lift sciences and given us a deep understanding of DNA sequences, RNA synthesis and the generation of proteins. This process can be represented as gene expression of molecular autoregulatory feedback loop systems. In this paper, the annealing robust fuzzy basis function (ARFBF) is proposed to improve the problems of fuzzy basis function for modeling of gene expression of molecular autoregulatory feedback loop systems with outliers. Firstly, the support vector regression (SVR) approach is proposed to determine the initial structure of ARFBF. Because of a SVR approach is equivalent to solving a linear constrained quadratic programming problem under a fixed structure of SVR, the number of hidden nodes, the initial parameters and the initial weights of ARFBF are easy obtained via the SVR approach. Secondly, the results of SVR are used as initial structure in ARFBF. At the same time, an annealing robust learning algorithm (ARLA) is used as the learning algorithm for ARFBF, and applied to adjust the parameters as well as weights of ARFBF. That is, an ARLA is proposed to overcome the problems of initialization and the cut-off points in the robust learning algorithm. Hence, when an initial structure of ARFBF is determined by a SVR approach, the ARFBF with ARLA have fast convergence speed and robust against outliers for the modeling of molecular autoregulatory feedback loop systems.

In this paper, the annealing robust fuzzy neural networks (ARFNNs) are proposed to improve the problems of fuzzy neural networks for modeling of molecular biology systems with outliers. Firstly, the support vector regression (SVR) approach is proposed to determine the initial structure of ARFNNs. Because of a SVR approach is equivalent to solving a linear constrained quadratic programming problem under a fixed structure of SVR the number of hidden nodes, the initial parameters and the initial weights of ARFNNs are easy obtained via the SVR approach. Secondly, the results of SVR are used as initial structure in ARFNNs. At the same time, an annealing robust learning algorithm (ARLA) is used as the learning algorithm for ARFNNs, and applied to adjust the parameters in the membership function as well as weights of ARFNNs. That is, an ARLA is proposed to overcome the problems of initialization and the cut-off points in the robust learning algorithm. Hence, when an initial structure of ARFNNs are determined by a SVR approach, the ARFNNs with ARLA have fast convergence speed and robust against outliers for the modeling of molecular biology systems.

This paper studies an intelligent system, including a support vector regression (SVR) and a similar analysis, for the classification of the ovarian cancer with microarray data. That is, steps in the classification include a feature selection step and a distance measure step. Firstly, the SVR is used to do the feature selection. That is, the SVR is applied to obtain the important genes of ovarian cancer for all samples of microarray data. At the same time, we can compute the frequency of the gene selection based on the results of SVR for all samples to determine the target genes of ovarian cancer. Secondly, the distance under the similar analysis between target data and test data can be determined. From the distance results, the classification of ovarian cancer can easy to determine

In this paper, an annealing robust Walsh function network (ARWFN) is proposed for modeling with outliers and its digital implementation. First, an annealing robust learning algorithm (ARLA) is used as the learning algorithm for the ARWFN, and applied to adjust the weights of ARWFNs. That is, an ARLA is proposed to overcome the problems of initialization and the cut-off points in the robust learning algorithm and deal with the model with noise and outliers. It turns out that the ARWFNs with ARLA present a fast convergence speed and are robust against outliers. Second, after the learning results, the ARWFNs are easy to implement using digital circuits. Simulation results are provided to show the validity and applicability of the proposed ARWFNs.

In this paper, a soft computing technique is proposed to select an initial structure of neuro-fuzzy networks (NFNs) for the noise data with outlier regression. That is, the proposed soft computing technique is fusion of neural network, fuzzy logic and support vector regression (SVR). Because the SVR approach is equivalent to solving a linear constrained quadratic programming problem under the fixed structure of SVR, the number of hidden nodes and adjustable parameters are easily obtained. That is, the SVR approach is used to obtain the initial structures of the NFNs in the noise data with outlier. Then, an annealing robust learning algorithm (ARLA) uses as the learning algorithm of NFNs. It is applied to adjust the parameters of promise parts and the fuzzy singleton of consequence parts in the NFNs. At the same time, the proposed method is applied to approximate a nonlinear model with outliers. Simulation results are provided to show the validity and applicability of the proposed soft computing technique.

The TSK type of fuzzy models has attracted a great attention of the fuzzy modeling community due to their good performance in various applications. Some approaches for modeling TSK fuzzy rules have been proposed in the literature. Most of them define their fuzzy subspaces bases based on the idea of training data being close enough instead of having similar functions. In addition, the fuzzy C-regression model (FCRM) clustering algorithm is proposed to construct TSK fuzzy models. However, this approach does not take into account the data distribution. In this paper, a novel TSK fuzzy modeling approach is presented. In this approach, adaptive fuzzy regression clustering (AFRC) algorithm is proposed to simultaneously define fuzzy subspaces and find the parameters in the consequent parts of TSK rules. In addition, the similarity measure is used to reduce the redundant rules in the clustering process. To obtain a more precise model, a gradient descent algorithm is employed. From the simulation results, the proposed TSK fuzzy model approach indeed showed superior performance.

In this paper, a new design procedure of nonlinear adaptive inverse control with the fuzzy neural networks based on support vector regression (FNNBSVR) and annealing robust learning algorithm (ARLA) is proposed for the magnetic bearing systems. The FNNBSVR is used to overcome initial structure problem and long training time in the nonlinear adaptive inverse control. Besides, the ARLA is proposed to overcome the outlier in the training procedure. It turns out that the proposed method can use less training time to get the FNNBSVR plant, inverse FNNBSVR plant and fuzzy neural network (FNN) controller to overcome the outlier in noise. Finally, this proposed method is applied to control magnetic bearing system. The experimental results show that the proposed method provides a greater flexibility and better performance in controlling magnetic bearing systems.

Proposes a new method to improve the gain scheduling technique for linear parameter-varying systems. We have derived the neural networks that can approximate the gain scheduling controller arbitrarily well when the sampling frequency satisfies the sample theorem. That means the gain scheduling controller can be fully reconstructed by neural networks using the discrete-time gain and time input as training pairs. Besides, we also show that the neural networks controller is independent of the sampling time. As a result, when the sampling rate is changed, the controller does not need to redesign. It turns out that the proposed neural networks controller has the following important properties: (i) the same performance as the continuous-time gain scheduling controller which can be generated by Algorithm 1 developed in this paper, (ii) less computing time than continuous-time gain scheduling controller (iii) good robustness against the sampling rates. Computer simulations show that the proposed method has better performance and less computing time than some existing gain scheduling techniques.

The authors propose a new neural structure with parallel and serial output via functional CPBUM neural network. Specifically, we combine the advantages in series to series and parallel to parallel to develop a parallel and serial output learning structure. Hence, the proposed model can reduce the computational complexity via the similarity analysis and sensitivity analysis. It is shown that the similarity analysis can be employed to determine the knowledge base of the controller. Furthermore, it is also shown that the sensitivity analysis can provide valuable information between input-output training pairs.

In this paper, we propose complex Chebyshev Polynomial Based unified model neural network for the approximation of complex- valued function. Based on this approximate transformable technique, we have derived the relationship between the single-layered neural network and multi-layered perceptron neural network. It is shown that the complex Chebyshev Polynomial Based unified model neural network can be represented as a functional link network that are based on Chebyshev polynomial. We also derived a new learning algorithm for the proposed network. It turns out that the complex Chebyshev Polynomial Based unified model neural network not only has the same capability of universal approximator, but also has faster learning speed than conventional complex feedforward/recurrent neural network.

In this paper, we propose a neural network model with a faster learning speed and a good approximate capability in the function approximation for solving worst-case identification of nonlinear systems H (infinity ) problems. Specifically, via the approximate transformable technique, we develop a Chebyshev Polynomials Based unified model neural network for solving the worst-case identification of nonlinear systems H (infinity ) problems. Based on this approximate transformable technique, the relationship between the single-layered neural network and multi-layered perceptron neural network is derived. It is shown that the Chebyshev Polynomials Based unified model neural network can be represented as a functional link network that is based on Chebyshev polynomials. We also derive a new learning algorithm such that the infinity norm of the transfer function from the input to the output is under a prescribed level. It turns out that the Chebyshev Polynomials Based unified model neural network not only has the same capability of universal approximator, but also has a faster learning speed than multi-layered perceptron or the recurrent neural network in the deterministic worst-case identification of nonlinear systems H (infinity ) problems.

In this paper, we propose the approximate transformable technique, which includes the direct transformation and indirect transformation, to obtain a CPB unified model neural networks for feedforward/recurrent neural networks via Chebyshev polynomials approximation. Based on this approximate transformable technique, we have derived the relationship between the single-layer neural networks and multilayer perceptron neural networks. It is shown that the CPB unified model neural networks can be represented as a functional link networks that are based on Chebyshev polynomials, and these networks use the recursive least squares method with forgetting factor as learning algorithm. It turns out that the CPB unified model neural networks not only has the same capability of universal approximator, but also has faster learning speed than conventional feedforward/recurrent neural networks. Computer simulations show that the proposed method does have the capability of universal approximator in some functional approximation with considerable reduction in learning time.