Deterministic Modelling of Randomness with Recurrent Artificial Neural Networks
It is shown that deterministic (chaotic) systems can be used to implicitly model the randomness of stochastic data, a question arising when addressing information processing in the brain according to the paradigm proposed by the EC APEREST project. More precisely, for a particular class of recurrent artificial neural networks, the identification procedure of stochastic signals leads to deterministic (chaotic) models which mimic the statistical/spectral properties of the original data.
KeywordsChaotic System Internal Node Chaos Synchronisation Chaotic Signal Chaotic Time Series
Unable to display preview. Download preview PDF.
- 1.De Feo, O., Hasler, M.: Modelling diversity by chaos and classification by synchronization. In: Pikovsky, A., Maistrenko, Y. (eds.) Synchronization: Theory and Application, pp. 25–40. Kluwer Academic Publishers, Dordrecht (2003)Google Scholar
- 7.Jaeger, H.: The echo state approach to analysing and training recurrent neural networks. Technical Report 148, Fraunhofer Institute for Autonomous Intelligent Systems (2001)Google Scholar
- 8.Baier, N.: Approximately Periodic Time Series and Nonlinear Structures. PhD thesis, Ecole Polytechnique Fédérale de Lausanne (2005)Google Scholar