Neural classifiers for dynamic system modes

doi:10.1016/0020-0255(94)00055-G

Information Sciences

Volume 82, Issues 3–4, January 1995, Pages 219-237

https://doi.org/10.1016/0020-0255(94)00055-G Get rights and content

Abstract

In this study we consider the use of neural networks for system identification. No a priori information is assumed about the system model. Rather it is assumed that the system has several modes of behavior and that a sequence of input-output pairs, while the system is in each mode, is available. Using the input-output information, a neural network is designed with the goal of recognizing each of the modes. The design is tested by using the input-output sequence of an unknown variable system, whose characteristics transition between those of the five prototypes. The identification procedure operates on-line. The ability to identify and track the transitions of the unknown system is explored.

References (22)

J.G. Proakis
Digital Communications
(1989)
D. Godard
Self-recovering equalization and carrier tracking in two-dimensional data communication systems
IEEE Trans. Commun.
(1980)
J.R. Treichler et al.
A new approach to multipath correction of constant modulus signals
IEEE Trans. Acoust., Speech, Signal Processing
(1983)
K.D. Kammeyer et al.
A modified adaptive FIR equalizer for multipath echo cancellation in FM transmission
IEEE J. Selected Areas Commun.
(1987)
P. Shalvi et al.
New criteria for blind deconvolution of nonminimum phase systems (channels)
IEEE Trans. Inform. Theory
(1990)
I.D. Landau
Adaptive Control, the Model Reference Approach
(1979)
G.C. Goodwin et al.
Adaptive Filtering, Prediction and Control
(1984)
L. Ljung et al.
Theory and Practice of Recursive Identification
(1983)
I.L. Leontaritis et al.
Input-output parametric models for non-linear systems. Part 1: Deterministic non-linear systems; Part 2: Stochastic non-linear systems
Int. J. Control
(1985)
Y.-H. Pao
Adaptive Pattern Recognition and Neural Networks
(1989)

S.R. Chu et al.

Neural network for system identification

IEEE Contr. Syst. Mag.

(1990)

Cited by (1)

Efficient exemplars for classifier design
1995, Journal of the Franklin Institute
Let X₁, …, X_m denote subsets of a vector space. A function ϕ is said to classify (distinguish, recognize, …) the sets X_j if the set images ϕ(X_j), j = 1, …, m are distinctive. Although a broad variety of classifier designs (statistical, separating hyperplane, neural, etc.) are available, a common problem plagues these designs whenever sets X_j are of high cardinality. Namely, the identification of exemplars which (i) are representative of the several classes, (ii) emphasize the importance of critical boundaries, and (iii) in total are of small cardinality. Such exemplar sets lead to robust designs, reduce computational costs, enhance algorithm convergence and often reduce the hardware attendant to design implementation.
In the present study we address the problem of choosing training sets for classifier designs. One result is an algorithm for selecting exemplars on the boundary of a set. Classifiers based on the boundary exemplars are effective for the binary problem: x ∈; X or x ∉ X. Using this algorithm we then present a procedure for obtaining boundary sets for the multiple classification case. These sets X′_j ⊂ X_j have the following properties: (i) the cardinality of X′_j is small, (ii) the critical boundaries between the X_j are delineated, and (iii) the sets X_j are synchronized in the sense that the boundary points are selected in pairs. These three properties suggest that the X′_j will facilitate a computational speed up in several classifier design methodologies, in particular nearest neighbor, separating hyperplane and neural network classifiers. We also present some preliminary tests which confirm this hypothesis.

View full text

Neural classifiers for dynamic system modes

Abstract

Digital Communications

Self-recovering equalization and carrier tracking in two-dimensional data communication systems

IEEE Trans. Commun.

A new approach to multipath correction of constant modulus signals

IEEE Trans. Acoust., Speech, Signal Processing

A modified adaptive FIR equalizer for multipath echo cancellation in FM transmission

IEEE J. Selected Areas Commun.

New criteria for blind deconvolution of nonminimum phase systems (channels)

IEEE Trans. Inform. Theory

Adaptive Control, the Model Reference Approach

Adaptive Filtering, Prediction and Control

Theory and Practice of Recursive Identification

Input-output parametric models for non-linear systems. Part 1: Deterministic non-linear systems; Part 2: Stochastic non-linear systems

Int. J. Control

Adaptive Pattern Recognition and Neural Networks

Neural network for system identification

IEEE Contr. Syst. Mag.