Preprocessing in Attractor Neural Networks

Abstract

Preprocessing the input patterns seems the simplest approach to invariant pattern recognition by neural networks. The Fourier transform has been proposed as an appropriate and elegant preprocessor. Nevertheless, we show in this work that the performance of this kind of preprocessor is strongly affected by the number of stored informations. This is because the phase of the Fourier transform plays a more important role than the amplitude in the recognition process.

to appear in Int.J.Mod.Phys.C, PostScript file