Eiji Uchino

Wavelet neural networks employing over-complete number of compactly supported non-orthogonal wavelets and their applications

This paper proposes two types of new neuron models, WS neuron (wavelet synapse neuron) and WA neuron (wavelet activation function neuron), which are obtained by modifying a traditional neuron model with non-orthogonal wavelet bases, while Boubez et al. (1993) employed orthonormal wavelets. Four types of typical wavelet neural networks employing WS and/or WA neurons are discussed. The simplest wavelet neural network exhibits much higher ability of generalization and much shorter time for learning rather than a three-layered feedforward neural network. Furthermore the wavelet neural network is guaranteed to give the global minimum. Other three wavelet neural networks are examined for predicting chaotic behaviour of a nonlinear dynamical system. The performance in learning speed and prediction of wavelet neural networks are more significant than a four-layered feedforward neural network.<<ETX>>

Soft Computing Based Signal Prediction, Restoration, and Filtering

In this chapter, soft computational signal processing, especially devoted to prediction, restoration and filtering of signals, is discussed. The neo-fuzzy-neuron, developed by the authors, are applied to the prediction and restoration of damaged signals. The chaotic signals and the speech signals are employed for the experiments. The filtering of noisy signals based on the Radial Basis Function (RBF) network, a special class of a fuzzy neural network, is also discussed. The proposed filter can eliminate not only Gaussian noise but also noise with an arbitrary distribution.

High speed fuzzy learning machine with guarantee of global minimum and its applications to chaotic system identification and medical image processing

The paper describes a generalized fuzzy learning machine, which is a generalised and modified type of the neo-fuzzy-neuron presented by the authors in 1992. This machine can well grasp the nonlinear correlation of each input. It has a very high nonlinear mapping ability compared with the conventional neural network and it guarantees the global minimum. Furthermore, learning speed and its accuracy are improved drastically. It was successfully applied to the identification of the nonlinear dynamical system, e.g. two dimensional Lorenz chaotic model, and to the automatic detection of landmark location in the roentgenographic cephalogram for orthodontic treatment. The results were promising.

Neo-fuzzy-neuron based new approach to system modeling, with application to actual system

This paper introduces a new approach to system modeling by using a neo-fuzzy-neuron. The system of concern is modeled adaptively by simply feeding to the neo-fuzzy-neuron, the basic principle of which was proposed by the authors in 1992, the input and the output data of the objective system. Firstly, the neo-fuzzy-neuron is applied to the restoration of a saturated and/or intermittent speech or chaotic signal to show its actual effectiveness. It is then extended in order to get a better generalization capability. An adaptive fuzzy modeling with use of a piece-wise linear membership function is also introduced. The experimental results have provided substantial proofs for their practical use.<<ETX>>

Fuzzy rule-based simple interpolation algorithm for discrete signal

Abstract This paper describes a fuzzy rule-based simple interpolation algorithm for discrete data. A simple interpolation algorithm between two noise-free points in the Euclidean space is proposed by taking into account the fuzzy effects of the surrounding points. The advanced algorithm for the noisy points is proposed, where interpolation is done by classifying the neighboring noisy points into some groups. The validity and the effectiveness of these methods have been verified by computer simulations and by applications to the actual time series noisy data. The experimental results have provided substantial proofs for practical use.

A design for a general digital filter for state estimation of an arbitrary stochastic sound system

This article describes a new attempt at the design of a general digital filter for the state estimation of a nonstationary nonlinear stochastic sound system. A recursive algorithm for estimating the higher-order statistics of arbitrary-function type, mean, and variance is obtained by introducing a new expansion form of Bayes theorem. Further, the state probability density function (PDF) can also be estimated in a unified form of orthogonal or nonorthogonal series expansions by using these estimates. This method is widely applicable for cases where the random-noise fluctuation is non-Gaussian. The estimation algorithm proposed in this article agrees completely with a well-known Kalman filtering theory [J. Basic Eng. 82, 35-45 (1960); Kalman and Buchy, J. Basic Eng. 83, 95-108 (1961)], as a simplified special case when the stochastic system is of linear type with Gaussian random excitation. The validity and effectiveness of the proposed theory were confirmed experimentally by applying it to actually observed room acoustic data and road-traffic noise data.

Nonlinear modeling and filtering by RBF network with application to noisy speech signal

Abstract This paper describes a nonlinear modeling and a filtering of an arbitrary nonlinear system based on a radial basis function (RBF) network. Modeling and filtering of the target system are performed by an RBF network. The modeling of a system and the synthesis of a filter are achieved by adjusting the connection weights of the RBF network by learning. The nonlinear filter proposed in this paper is widely applicable to elimination of noise with an arbitrary distribution. The effectiveness and the validity of the present method have been confirmed by applying it to the modeling and the filtering of a noisy speech signal.

Hardware implementation of fuzzy filtering

The authors describe a simple interpolation algorithm for noisy signal data using fuzzy inference and its hardware implementation. Concretely, keeping in mind that even if the signal data are disturbed by noise, a rough sketch of the true signal pattern can be generally made, an interpolation algorithm based on fuzzy logic is proposed. It results eventually in the reduction of noise in the signal data with no knowledge of its dynamics. The effectiveness of the method was verified by computer simulations, and the method was implemented by a hybrid electronic circuit on a breadboard.<<ETX>>

An approach to designing the fuzzy IF-THEN rules for fuzzy-controlled static var compensator (FCSVC)

Abstract This paper describes an approach to designing fuzzy if-then rules for the fuzzy-controlled static var compensator (FCSVC) in a three-phase electric power system. In general, it is very difficult to control the rms line voltage in the three-phase ac circuit. We propose the FCSVC system to control the voltage. FCSVC is an rms line voltage stabilizer using three static var compensators (SVC) which are controlled by a fuzzy logic controller (FLC). Moreover, we propose an easier and more efficient approach to designing the fuzzy if-then rules of FCSVC. The effectiveness of the FCSVC described in this paper is verified by the experimental results.

Behavior of chaotic network hardware system

This paper describes the behavior of the chaotic unit employing the chaotic chip and that of the chaotic network. The experimental results were discussed with computer simulation results. The chip was developed for analyzing nonlinear dynamical network systems. It operates in analog voltage mode with switched capacitors and clock signal and generates discrete time series signal. It includes a 3-segment piecewise linear mapping function, the nonlinear parameters of which can be assigned externally.<<ETX>>