Function approximation using fuzzy neural networks with robust learning algorithm

Preliminary study on wilcoxon learning machines

As is well known in statistics, the resulting linear regressors by using the rank-based Wilcoxon approach to linear regression problems are usually robust against (or insensitive to) outliers. This motivates us to introduce in this paper the Wilcoxon approach to the area of machine learning. Specifically, we investigate four new learning machines, namely Wilcoxon neural network (WNN), Wilcoxon generalized radial basis function network (WGRBFN), Wilcoxon fuzzy neural network (WFNN), and kernel-based Wilcoxon regressor (KWR). These provide alternative learning machines when faced with general nonlinear learning problems. Simple weights updating rules based on gradient descent will be derived. Some numerical examples will be provided to compare the robustness against outliers for various learning machines. Simulation results show that the Wilcoxon learning machines proposed in this paper have good robustness against outliers. We firmly believe that the Wilcoxon approach will provide a promising methodology for many machine learning problems.

Dynamic optimal learning rates of a certain class of fuzzy neural networks and its applications with genetic algorithm

The stability analysis of the learning rate for a two-layer neural network (NN) is discussed first by minimizing the total squared error between the actual and desired outputs for a set of training vectors. The stable and optimal learning rate, in the sense of maximum error reduction, for each iteration in the training (back propagation) process can therefore be found for this two-layer NN. It has also been proven in this paper that the dynamic stable learning rate for this two-layer NN must be greater than zero. Thus it Is guaranteed that the maximum error reduction can be achieved by choosing the optimal learning rate for the next training iteration. A dynamic fuzzy neural network (FNN) that consists of the fuzzy linguistic process as the premise part and the two-layer NN as the consequence part is then illustrated as an immediate application of our approach. Each part of this dynamic FNN has its own learning rate for training purpose. A genetic algorithm is designed to allow a more efficient tuning process of the two learning rates of the FNN. The objective of the genetic algorithm is to reduce the searching time by searching for only one learning rate, which is the learning rate of the premise part, in the FNN. The dynamic optimal learning rates of the two-layer NN can be found directly using our innovative approach. Several examples are fully illustrated and excellent results are obtained for the model car backing up problem and the identification of nonlinear first order and second order systems.

Handling uncertainties in toxicity modelling using a fuzzy filter

A fundamental concern in the Quantitative Structure-Activity Relationship approach to toxicity evaluation is the generalization of the model over a wide range of compounds. The data driven modelling of toxicity, due to the complex and ill-defined nature of eco-toxicological systems, is an uncertain process. The development of a toxicity predicting model without considering uncertainties may produce a model with a low generalization performance. This study presents a novel approach to toxicity modelling that handles the involved uncertainties using a fuzzy filter, and thus improves the generalization capability of the model. The method is illustrated by considering a data set dealing with the fathead minnow (Pimephales promelas) toxicity of 568 organic compounds.

Deterministic approach to robust adaptive learning of fuzzy models

This study is concerned with the adaptive learning of an interpretable Sugeno-type fuzzy inference system, in a deterministic framework, in the presence of data uncertainties and modeling errors. The authors explore the use of Hinfinity estimation theory and least squares estimation for online learning of membership functions and consequent parameters without making any assumption and requiring a priori knowledge of upper bounds, statistics, and distribution of data uncertainties and modeling errors. The issues of data uncertainties, modeling errors, and time variations have been considered mathematically in a sensible way. The proposed robust approach to the adaptive learning of fuzzy models has been illustrated through the examples of adaptive system identification, time-series prediction, and estimation of an uncertain process.

Robust and fast learning for fuzzy cerebellar model articulation controllers

In this paper, the online learning capability and the robust property for the learning algorithms of cerebellar model articulation controllers (CMAC) are discussed. Both the traditional CMAC and fuzzy CMAC are considered. In the study, we find a way of embeding the idea of M-estimators into the CMAC learning algorithms to provide the robust property against outliers existing in training data. An annealing schedule is also adopted for the learning constant to fulfill robust learning. In the study, we also extend our previous work of adopting the credit assignment idea into CMAC learning to provide fast learning for fuzzy CMAC. From demonstrated examples, it is clearly evident that the proposed algorithm indeed has faster and more robust learning. In our study, we then employ the proposed CMAC for an online learning control scheme used in the literature. In the implementation, we also propose to use a tuning parameter instead of a fixed constant to achieve both online learning and fine-tuning effects. The simulation results indeed show the effectiveness of the proposed approaches.

An Online GA-Based Output-Feedback Direct Adaptive Fuzzy-Neural Controller for Uncertain Nonlinear Systems

In this paper, we propose a novel design of a GA-based output-feedback direct adaptive fuzzy-neural controller (GODAF controller) for uncertain nonlinear dynamical systems. The weighting factors of the direct adaptive fuzzy-neural controller can successfully be tuned online via a GA approach. Because of the capability of genetic algorithms (GAs) in directed random search for global optimization, one is used to evolutionarily obtain the optimal weighting factors for the fuzzy-neural network. Specifically, we use a reduced-form genetic algorithm (RGA) to adjust the weightings of the fuzzy-neural network. In RGA, a sequential-search -based crossover point (SSCP) method determines a suitable crossover point before a single gene crossover actually takes place so that the speed of searching for an optimal weighting vector of the fuzzy-neural network can be improved. A new fitness function for online tuning the weighting vector of the fuzzy-neural controller is established by the Lyapunov design approach. A supervisory controller is incorporated into the GODAF controller to guarantee the stability of the closed-loop nonlinear system. Examples of nonlinear systems controlled by the GODAF controller are demonstrated to illustrate the effectiveness of the proposed method.