A normalised backpropagation learning algorithm for multilayer feed-forward neural adaptive filters

A. I. Hanna; D. P. Mandic; M. Razaz

doi:10.1109/NNSP.2001.943111

A normalised backpropagation learning algorithm for multilayer feed-forward neural adaptive filters

A. I. Hanna
, D. P. Mandic
, M. Razaz

School of Computing Sciences

Research output: Contribution to conference › Paper

9 Citations (Scopus)

Abstract

Analysis of a normalised backpropagation (NBP) algorithm employed in feed-forward multilayer nonlinear adaptive filters trained by backpropagation is provided. It is first shown that a degree of freedom in training of a nonlinear adaptive filter can be removed according to the relationship between the gain of the activation function, learning rate and weight matrix. The derivation of the NBP algorithm for a multilayer feed-forward neural adaptive filter is then provided based upon the minimisation of the instantaneous output error of the filter. Simulation results show that the NBP algorithm converges faster than a standard backpropagation algorithm and achieves better prediction gain when applied to nonlinear and non-stationary signals

Original language	English
Pages	63-72
Number of pages	10
DOIs	https://doi.org/10.1109/NNSP.2001.943111
Publication status	Published - Sept 2001
Event	2001 IEEE Signal Processing Society Workshop on Neural Nets for Signal Processing XI - North Falmouth, United States Duration: 10 Sept 2001 → 12 Sept 2001

Conference

Conference	2001 IEEE Signal Processing Society Workshop on Neural Nets for Signal Processing XI
Country/Territory	United States
City	North Falmouth
Period	10/09/01 → 12/09/01

Access to Document

10.1109/NNSP.2001.943111

Cite this

@conference{fe9680fae23f456cba7202e5666014b2,

title = "A normalised backpropagation learning algorithm for multilayer feed-forward neural adaptive filters",

abstract = "Analysis of a normalised backpropagation (NBP) algorithm employed in feed-forward multilayer nonlinear adaptive filters trained by backpropagation is provided. It is first shown that a degree of freedom in training of a nonlinear adaptive filter can be removed according to the relationship between the gain of the activation function, learning rate and weight matrix. The derivation of the NBP algorithm for a multilayer feed-forward neural adaptive filter is then provided based upon the minimisation of the instantaneous output error of the filter. Simulation results show that the NBP algorithm converges faster than a standard backpropagation algorithm and achieves better prediction gain when applied to nonlinear and non-stationary signals",

author = "Hanna, \{A. I.\} and Mandic, \{D. P.\} and M. Razaz",

year = "2001",

month = sep,

doi = "10.1109/NNSP.2001.943111",

language = "English",

pages = "63--72",

note = "2001 IEEE Signal Processing Society Workshop on Neural Nets for Signal Processing XI ; Conference date: 10-09-2001 Through 12-09-2001",

}

TY - CONF

T1 - A normalised backpropagation learning algorithm for multilayer feed-forward neural adaptive filters

AU - Hanna, A. I.

AU - Mandic, D. P.

AU - Razaz, M.

PY - 2001/9

Y1 - 2001/9

N2 - Analysis of a normalised backpropagation (NBP) algorithm employed in feed-forward multilayer nonlinear adaptive filters trained by backpropagation is provided. It is first shown that a degree of freedom in training of a nonlinear adaptive filter can be removed according to the relationship between the gain of the activation function, learning rate and weight matrix. The derivation of the NBP algorithm for a multilayer feed-forward neural adaptive filter is then provided based upon the minimisation of the instantaneous output error of the filter. Simulation results show that the NBP algorithm converges faster than a standard backpropagation algorithm and achieves better prediction gain when applied to nonlinear and non-stationary signals

AB - Analysis of a normalised backpropagation (NBP) algorithm employed in feed-forward multilayer nonlinear adaptive filters trained by backpropagation is provided. It is first shown that a degree of freedom in training of a nonlinear adaptive filter can be removed according to the relationship between the gain of the activation function, learning rate and weight matrix. The derivation of the NBP algorithm for a multilayer feed-forward neural adaptive filter is then provided based upon the minimisation of the instantaneous output error of the filter. Simulation results show that the NBP algorithm converges faster than a standard backpropagation algorithm and achieves better prediction gain when applied to nonlinear and non-stationary signals

U2 - 10.1109/NNSP.2001.943111

DO - 10.1109/NNSP.2001.943111

M3 - Paper

SP - 63

EP - 72

T2 - 2001 IEEE Signal Processing Society Workshop on Neural Nets for Signal Processing XI

Y2 - 10 September 2001 through 12 September 2001

ER -