The usefulness of input preprocessing for feedforward neural architecture is presented and discussed. We have shown importance of signal normalization, choice of sampling time, nonlinear transformation to make input signals spectra compatible and dynamical orthonormal preprocessing for non-static systems. In particular some serious limitations of widely know tapped delay architecture are presented.
CITATION STYLE
Beliczynski, B. (2004). Input preprocessing for neural feedforward architecture. In Intelligent Information Processing and Web Mining (pp. 633–641). Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-540-39985-8_71
Mendeley helps you to discover research relevant for your work.