Robust Learning of Chaotic Attractors

Abstract

A fundamental problem with the modeling of chaotic time series data is that minimizing short-term prediction errors does not guarantee a match between the reconstructed attractors of model and experiments. We introduce a modeling paradigm that simultaneously learns to short-tenn predict and to locate the outlines of the attractor by a new way of nonlinear principal component analysis. Closed-loop predictions are constrained to stay within these outlines, to prevent divergence from the attractor. Learning is exceptionally fast: parameter estimation for the 1000 sample laser data from the 1991 Santa Fe time series competition took less than a minute on a 166 MHz Pentium PC.

Cite

Text

Bakker et al. "Robust Learning of Chaotic Attractors." Neural Information Processing Systems, 1999.

Markdown

[Bakker et al. "Robust Learning of Chaotic Attractors." Neural Information Processing Systems, 1999.](https://mlanthology.org/neurips/1999/bakker1999neurips-robust/)

BibTeX

@inproceedings{bakker1999neurips-robust,
  title     = {{Robust Learning of Chaotic Attractors}},
  author    = {Bakker, Rembrandt and Schouten, Jaap C. and Coppens, Marc-Olivier and Takens, Floris and Giles, C. Lee and van den Bleek, Cor M.},
  booktitle = {Neural Information Processing Systems},
  year      = {1999},
  pages     = {879-885},
  url       = {https://mlanthology.org/neurips/1999/bakker1999neurips-robust/}
}