Learning and predicting time series by neural networks
Disordered Systems and Neural Networks
2009-11-07 v1
Abstract
Artificial neural networks which are trained on a time series are supposed to achieve two abilities: firstly to predict the series many time steps ahead and secondly to learn the rule which has produced the series. It is shown that prediction and learning are not necessarily related to each other. Chaotic sequences can be learned but not predicted while quasiperiodic sequences can be well predicted but not learned.
Keywords
Cite
@article{arxiv.cond-mat/0202537,
title = {Learning and predicting time series by neural networks},
author = {Ansgar Freking and Wolfgang Kinzel and Ido Kanter},
journal= {arXiv preprint arXiv:cond-mat/0202537},
year = {2009}
}
Comments
5 pages