Neural Network analysis of time series
R. A. Calvo, H. Navone, H. A. Ceccatto
Neural Networks have proven to be usefull tools in time series
analysis and forecasting. For those researchers considering using
these techniques, we list a series of resources that can be helpull.
Software
You will find in this monthly posting to the
Usenet newsgroup comp.ai.neural-nets a very complete list of
freeware, shareware and comercial products.
We can not suggest you any of these, since each of them have different
features that you might find attractive/usefull. After clarifying
this, we can say that we found that
Stuttgart Neural Network Simulator (SNNS) , satisfied a number of constraints:
Free: we didn't want to advertise a particular comercial product.
Active: many of freeware packages do not have an active group of
researchers using it, so no support at all can be expected. This one does.
Documentation: It is essential, that whatever package you choose
have a complete user manual. SNNS provides you of a book, with
examples, some background information of the theory and a complete
description of the implementation.
Data
Tools
You might find usefull some of these scripts. You only need a perl interpreter installed in your machine.
Links
(berkeley): Learning
(statistics, neural networks, discovery, theory) Software environments for
learning Prototyping tools for learning Probabilistic modeling Unification
and applied theory for the above.
Todd Leen (Oregon Graduate Institute)
Stochastic and deterministic learning dynamics, Self-organizing models
for representation of high-D data, Locally linear modeling techniques
, Model pruning and generalization, Invariance in learning
Rafael A Calvo
Last modified: Thu Aug 20 19:17:48 EDT 1998