Presentation 1996/10/28
Layerd Neural Networks with Autoregressive Moving Average Links
T. KAMIYAMA, T. FURUYA, M. SEKIGUCHI, T. TANAKA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a neural net(ARMANN) whose link has an autoregressive moving average(ARMA) model. The ARMA model consists of an sutoregerssive(AR) model that works as a delayed feedback and a moving average(MA) model that works as a delayed feedforward. The neural network with ARMA models can process temporal data by storing the past data in its delay elements. We present a learning algorithm of ARMANN based on the packpropagation. ARMANN is applied to temporal data processing problem (abcde trajectory, stock market prediction, inverse model). The results show that the ARMANN is superior to ARLNN in the construction of inverse model.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Layerd neural network / Autoregressive moving average model / Backpropagation
Paper # NC96-43
Date of Issue

Conference Information
Committee NC
Conference Date 1996/10/28(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 JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Layerd Neural Networks with Autoregressive Moving Average Links
Sub Title (in English)
Keyword(1) Layerd neural network
Keyword(2) Autoregressive moving average model
Keyword(3) Backpropagation
1st Author's Name T. KAMIYAMA
1st Author's Affiliation Toho University()
2nd Author's Name T. FURUYA
2nd Author's Affiliation Toho University
3rd Author's Name M. SEKIGUCHI
3rd Author's Affiliation Toho University
4th Author's Name T. TANAKA
4th Author's Affiliation Electrotechnical Laboratory
Date 1996/10/28
Paper # NC96-43
Volume (vol) vol.96
Number (no) 331
Page pp.pp.-
#Pages 8
Date of Issue