We address the important theoretical question why a recurrent neural network with fixed weights can adaptively classify time-varied signals in the presence of additive noise and parametric perturbations. We provide a mathematical proof assuming that unknown parameters are allowed to enter the signal nonlinearly and the noise amplitude is sufficiently small.
@article{arxiv.0705.3370,
title = {Adaptive classification of temporal signals in fixed-weights recurrent neural networks: an existence proof},
author = {Ivan Tyukin and Danil Prokhorov and Cees van Leeuwen},
journal= {arXiv preprint arXiv:0705.3370},
year = {2007}
}