This paper presents the architecture and learning procedure underlying ANFIS (AdaptiveNetwork -based Fuzzy Inference System), a fuzzy inference system implemented in the framework of adaptive networks. By using a hybrid learning procedure, the proposed ANFIS can construct an input-output mapping based on both human knowledge (in the form of fuzzy if-then rules) and stipulated input-output data pairs. In our simulation, we employ the ANFIS architecture to model nonlinear functions, identify nonlinear components on-linely in a control system, and predict a chaotic time series, all yielding remarkable results. Comparisons with artificail neural networks and earlier work on fuzzy modeling are listed and discussed. Other extensions of the proposed ANFIS and promising applications to automatic control and signal processing are also suggested. 1 Introduction System modeling based on conventional mathematical tools (e.g., differential equations) is not well suited for dealing with ill-define...

Neural networks are composed of basic units somewhat analogous to neurons. These units are linked to each other by connections whose strength is modifiable as a result of a learning process or algorithm. Each of these units integrates independently (in parallel) the information provided by its synapses in order to evaluate its state of activation. The unit response is then a linear or nonlinear function of its activation. Linear algebra concepts are used, in general, to analyze linear units, with eigenvectors and eigenvalues being the core concepts involved. This analysis makes clear the strong similarity between linear neural networks and the general linear model developed by statisticians. The linear models presented here are the perceptron, and the linear associator. The behavior of nonlinear networks can be described within the framework of optimization and approximation techniques with dynamical systems (e.g., like those used to model spin glasses). One of the main notio...

In this paper we develop a stochastic realization theory for multiscale autoregressive (MAR) processes that leads to computationally efficient realization algorithms. The utility of MAR processes has been limited by the fact that the previously known general purpose realization algorithm, based on canonical correlations, leads to model inconsistencies and has complexity quartic in problem size. Our realization theory and algorithms addresses these issues by focusing on the estimation-theoretic concept of predictive efficiency and by exploiting the scale-recursive structure of so-called internal MAR processes. Our realization algorithm has complexity quadratic in problem size and with an approximation we also obtain an algorithm that has complexity linear in problem size.

By combining methods from artificial intelligence and signal analysis, we have developed a hybrid system for medical diagnosis. The core of the system is a fuzzy expert system with a dual source knowledge base. Two sets of rules are acquired, inductively from given examples and deductively formulated by the physician. A fuzzy neural network serves to learn from sample data and allows to extract fuzzy rules for the knowledge base. A complex signal transformation preprocesses the digital data a priori to the symbolic representation. Results demonstrate the high accuracy of the system in the field of diagnosing electroencephalograms where it outperforms the visual diagnosis by a human expert for some phenomena. KEYWORDS: Expert Systems, Fuzzy Logic, Hybrid Systems, Medical Diagnosis, Neural Networks 1 Introduction The emerging need to evaluate a vast variety of electronic patient data, often in the form of multidimensional signals, raises the demand for automated diagnosis methods. We ha...

Vehicle acoustic signals have long been considered as unwanted traffic noise. In this research acoustic signals generated by each vehicle will be used to detect its presence and classify its type. Circular arrays of microphones were designed and built to detect desired signals and suppress unwanted ones. Circular arrays with multiple rings have an interesting and important property that is constant sidelobe levels. A modified genetic algorithm that can work directly with real numbers is used in the circular array design. It offers more effective ways to solve numerical problems than a standard genetic algorithm.

This paper presents a novel approach to the segmentation and integration of #radar# images using a secondorder

A 1-dimensional wavelet transform is a method of expansion of a single-variable function into a combination of generic functions called "wavelets". Wavelets are generated from a single appropriately selected function by operations of dilation and translation. Expansion into wavelets captures the essential time-frequency properties of a function. Recently, the 2-dimensional wavelet transform has found an application in image coding. The 2-D wavelet transform followed by vector quantization gives a possibility to encode the image data with a low bit rate without significant loss in quality. This explains the growing popularity of the wavelet transform. In this report we will cover the theoretical foundation of the wavelet transform and present its application in image coding. The problem of optimal bit allocation for quantization of wavelet coefficients will be also examined. The report will be concluded with some experimental results of this image coding. Key Words: Image Coding, Wavel...

II. Classification and identification Artificial neural networks as a tool for species identification of fish schools

In this paper, the authors have attempted the identification of linear time-varying discrete time dynamic system. It is supposed that such systems are signified by transfer function characterizations. As the behaviour of the system changes, the neuron model developed keeps track of the changes in the characteristics and parameters of the system. Thus, at any instant of time, it correctly simulates the given time-varying system, despite the significant changes in system’s property. The excellent approximating capability of the neural network is used to identify the relationship between system variables and parameters. In essence, a neural network perfectly mimics and identifies the actual physical system. It is shown that a simple feedforward neural network containing a single neuron fairly accurately simulates the linear dynamical time-variant system under consideration, which may have Auto-regressive or Moving-average or Autoregressive moving-average model. Keywords: Dynamical; Time-variant; Auto regressive model; Moving average model; Auto regressive moving average model with exogenous inputs; Adaptive system identification; Neural network

i Aggregation of the "proper" input signals results in the activation potential, v, which can be expressed as the inner product of "proper" input signals and related weights: v = p 1 X i=1 w i x i = w 1:p 1 x 1:p 1 A.P.Paplinski 3--1 NNets --- L. 3 March 17, 1999 The augmented activation potential, ^ v, can now be expressed simply as: ^ v = w x = v For each input signal, the output is determined as y = (^v) = 8 < : 0 if v < (^v < 0) 1 if v <F73.8