In this article we examine how model selection in neural networks can be guided by statistical procedures such as hypotheses tests, information criteria and cross validation. The application of these methods in neural network models is discussed, paying attention especially to the identification...

A general model specification test of a parametric model against a nonparametric or semiparametric alternative is studied. The test statistic employs a fixed kernel, not varying by a bandwidth. This test is proved to be consistent, the asymptotic distribution is derived and shown to be...

Most standard hypothesis tests have high power only against a limited space of alternative hypotheses. With the advent of new tests for the same hypothesis, claimed to have higher power for some alternatives, but lower for other, the practitioner has to make a choice between alternative tests....