Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics
Abstract
We present a method for the unsupervised segmentation of data streams originating from different unknown sources that alternate in time. We use an architecture consisting of competing neural networks. Memory is included to resolve ambiguities of input-output relations. To obtain maximal specialization, the competition is adiabatically increased during training. Our method achieves almost perfect identification and segmentation in the case of switching chaotic dynamics where input manifolds overlap and input-output relations are ambiguous. Only a small dataset is needed for the training procedure. Applications to time series from complex systems demonstrate the potential relevance of our approach for time series analysis and short-term prediction.
Cite
Text
Pawelzik et al. "Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics." Neural Computation, 1996. doi:10.1162/NECO.1996.8.2.340Markdown
[Pawelzik et al. "Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics." Neural Computation, 1996.](https://mlanthology.org/neco/1996/pawelzik1996neco-annealed/) doi:10.1162/NECO.1996.8.2.340BibTeX
@article{pawelzik1996neco-annealed,
title = {{Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics}},
author = {Pawelzik, Klaus and Kohlmorgen, Jens and Müller, Klaus-Robert},
journal = {Neural Computation},
year = {1996},
pages = {340-356},
doi = {10.1162/NECO.1996.8.2.340},
volume = {8},
url = {https://mlanthology.org/neco/1996/pawelzik1996neco-annealed/}
}