Back to Search
Start Over
Stability Analysis of Deep Belief Network: Based SD-AR Model for Nonlinear Time Series.
- Source :
- Neural Processing Letters; Apr2024, Vol. 56 Issue 2, p1-20, 20p
- Publication Year :
- 2024
-
Abstract
- As for nonlinear time series prediction, many different kinds of varying-coefficient models have been proposed and analysised in recent years. A kind of varying functional-coefficient autoregressive model, called the deep belief network-based state-dependent autoregressive (DBN-AR) model is considered in this paper. The stability conditions and existing conditions of limit cycle of the DBN-AR model are also studied. An especial designed parameter estimation method is used to identify the DBN-AR model. The DBN-AR model is used to predict the famous Canadian lynx data and Henon chaotic series, the prediction capability of the DBN-AR model is compared with other prediction models, the experimental results show that the DBN-AR model obtains better prediction accuracy. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 13704621
- Volume :
- 56
- Issue :
- 2
- Database :
- Complementary Index
- Journal :
- Neural Processing Letters
- Publication Type :
- Academic Journal
- Accession number :
- 175622760
- Full Text :
- https://doi.org/10.1007/s11063-024-11543-x