Presentation 1997/5/23
An Input Sequence Analysis for a Hybrid Neural Predictor
Ashraf A.M. Khalaf, Kenji Nakayama,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In order to measure the difficulty of nonlinear time sereis prediction, we propose some measure, with which correlation among the input samples is evaluated. Furthermore, relation between the number of the input samples, applied to the predictor in parallel, and prediction performance can be estimated based on this measure. A cascade structure of multi-layer (ML) neural network combined with a finite-impulse-response (FIR) filter is suggested for one-step prediction of nonlinear time series. Using the FIR as a second stage of the predictor speeds up the convergence of the proposed network for different input samples and hidden neurons combinations.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Nonlinear Signal Processing / Neural Networks / FIR Filter / Hybrid Predictor Models
Paper # NC97-2
Date of Issue

Conference Information
Committee NC
Conference Date 1997/5/23(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) An Input Sequence Analysis for a Hybrid Neural Predictor
Sub Title (in English)
Keyword(1) Nonlinear Signal Processing
Keyword(2) Neural Networks
Keyword(3) FIR Filter
Keyword(4) Hybrid Predictor Models
1st Author's Name Ashraf A.M. Khalaf
1st Author's Affiliation Graduate School of Nat.Sci.and Tech.()
2nd Author's Name Kenji Nakayama
2nd Author's Affiliation Faculty of Eng., Kanazawa Univ.
Date 1997/5/23
Paper # NC97-2
Volume (vol) vol.97
Number (no) 69
Page pp.pp.-
#Pages 8
Date of Issue