Modelling nonstationary dynamics
✍ Scribed by M.I. Széliga; P.F. Verdes; P.M. Granitto; H.A. Ceccatto
- Publisher
- Elsevier Science
- Year
- 2003
- Tongue
- English
- Weight
- 113 KB
- Volume
- 327
- Category
- Article
- ISSN
- 0378-4371
No coin nor oath required. For personal study only.
✦ Synopsis
We incorporate the use of validation data to cope with noisy records in a neural network-based method for modelling the dynamics of slowly changing nonstationary systems. As a byproduct, we obtain a precise criterion to ÿnd the optimal value of a required internal hyperparameter. Testing these ideas on a controlled problem shows that the resulting algorithm is able to outperform previous methods in the literature, allowing a more accurate modelling of nonstationary dynamics.
📜 SIMILAR VOLUMES
Stochastic processes observed over space often exhibit nonstationarity. Possible causes of nonstationarity include mean drift, heterogeneity of responses, or a correlation pattern that is not simply a function of the Euclidean distance between two spatial locations. This paper considers the latter.
Most temporal signals of practical interest are nonstationary and need to be modeled using time-varying systems. In this paper a composite model for these signals is proposed which accounts for nonstationarities in the mean and the autocorrelation functions. Unbiased and consistent time-varying esti