The nonparametric detector using neural network

The authors point out that there is no distinction between the terms `non-parametric´ and `distribution-free´ in most engineering applications. They show the non-parametric nature of neural networks (NNs) on two aspects of theory and experiment, and conclude that a NN-detector is a kind of non-parametric detector. The basis of the theoretical derivation is the Lyapunov theorem in probability statistics. The experimental studies, with the introduction of noise with Gaussian, Rayleigh, exponential, and uniform distribution, yielded good results. The NN-detector described not only gives a kind of non-parametric detector with a new structure, but also provides a new systematic design method for the non-parametric detector