The Observer-Observation Dilemma in Neuro-Forecasting
Abstract
We explain how the training data can be separated into clean informa(cid:173) tion and unexplainable noise. Analogous to the data, the neural network is separated into a time invariant structure used for forecasting, and a noisy part. We propose a unified theory connecting the optimization al(cid:173) gorithms for cleaning and learning together with algorithms that control the data noise and the parameter noise. The combined algorithm allows a data-driven local control of the liability of the network parameters and therefore an improvement in generalization. The approach is proven to be very useful at the task of forecasting the German bond market.
Cite
Text
Zimmermann and Neuneier. "The Observer-Observation Dilemma in Neuro-Forecasting." Neural Information Processing Systems, 1997.Markdown
[Zimmermann and Neuneier. "The Observer-Observation Dilemma in Neuro-Forecasting." Neural Information Processing Systems, 1997.](https://mlanthology.org/neurips/1997/zimmermann1997neurips-observerobservation/)BibTeX
@inproceedings{zimmermann1997neurips-observerobservation,
title = {{The Observer-Observation Dilemma in Neuro-Forecasting}},
author = {Zimmermann, Hans-Georg and Neuneier, Ralph},
booktitle = {Neural Information Processing Systems},
year = {1997},
pages = {992-998},
url = {https://mlanthology.org/neurips/1997/zimmermann1997neurips-observerobservation/}
}