A Neural Network Based Nonlinear Temporal-Spatial Noise Rejection System

摘要

This paper proposes a nonlinear temporal-spatial noise rejection system on the basis of mapping neural networks. With the universe nonlinear mapping capability of these neural networks and related learning algorithms, the proposed system can offer better noise rejection performance than traditional methods in the case that the related unknown system is nonlinear or non-minimum phase and in the case that the length of the learning system does not fit the length of the unknown system. It can then serve as an alternative tool for many applications of noise rejection and this was confirmed by the simulations results.