International Symposium on Nonlinear Theory and its Applications
Estimation of Embedding Dimension Using Self-Organizing Map
Haruna MATSUSHITA, Yoshifumi NISIHO,
PDF download (1MB)
In order to analyze the nonlinear time series, the embedding is very important. In the em- bedding process, we must determine the appropriate embedding dimension to reconstruct with time-delay. In this study, we propose a method to estimate the em- bedding dimension using Self-Organizing Map (SOM). We can obtain the map re?ecting the distribution state of input data using SOM. We carry out simulations for the time series obtained from the Henon map, the Ikeda map, and the Lorenz equations, and con?rm that our method would be useful to estimate the appropri- ate embedding dimensions.