Effect of time-correlation of input patterns on the convergence of on-line learning
adap-org
2009-10-28 v1 Adaptation and Self-Organizing Systems
Abstract
We studied the effects of time correlation of subsequent patterns on the convergence of on-line learning by a feedforward neural network with backpropagation algorithm. By using chaotic time series as sequences of correlated patterns, we found that the unexpected scaling of converging time with learning parameter emerges when time-correlated patterns accelerate learning process.
Cite
@article{arxiv.adap-org/9606005,
title = {Effect of time-correlation of input patterns on the convergence of on-line learning},
author = {Tsuyoshi Hondou and Mitsuaki Yamamoto and Yasuji Sawada and Yoshihiro Hayakawa},
journal= {arXiv preprint arXiv:adap-org/9606005},
year = {2009}
}
Comments
8 pages(Revtex), 5 figures