Parallel Performance of Ensemble Self-Generating Neural Networks for Chaotic Time Series Prediction Problems

スポンサーリンク

概要

• 論文の詳細を見る

• Self-Generating Neural Network (SGNN) have a feature of the fast processing by automatically constructing self-generating neural tree (SGNT) from the given training data set. The prediction accuracy of SGNN for chaotic time series prediction is improved by the ensemble averaging of various SGNN. However, the computation time commensurately increases the number of SGNN on a single processor. In this paper, we investigate the improving capability of the prediction accuracy and the parallel efficiency to ensemble SGNN for chaotic time series prediction problems on a MIMD parallel computer. We allocate each SGNN to each processor. Our results show that the more the number of node processors increase, the more the improvement of the prediction accuracy is obtained for all problems, and keep the characteristic of the high speed processing of the single SGNN.

• 社団法人電子情報通信学会の論文
• 2002-02-01

著者

• Inoue Hirotaka

  Graduate School Of Engineering Okayama University Of Science

• Inoue Hirotaka

  Graduate School Of Agricultural And Life Sciences The University Of Tokyo

• Narihisa Hiroyuki

  Department Of Information & Computer Engineering Faculty Of Engineering Okayama University Of Sc

• INOUE Hirotaka

  Graduate School of Engineering, Okayama University of Science

関連論文

• Calcium-binding Abilities of Two Cuticle Peptides from the Exoskeleton of the Crayfish, Procambarus clarkii
• APPROXIMATE METHOD FOR SEPARABLE NONCONVEX PROGRAMS
• Parallel Performance of Ensemble Self-Generating Neural Networks for Chaotic Time Series Prediction Problems
• An Efficiency Enhancement of k-opt Local Search Heuristic for Binary Quadratic Programming Problem
• Efficient Hybrid Neural Network for Chaotic Time Series Prediction

スポンサーリンク