Introduction to Special Issue

Abstract

In recent years, neural-based techniques have been proved effective in solving several applications by outperforming, whenever the world is non-linear and unspecified, traditional algorithmic-based approaches. In fact, if the equations ruling a generic application are unknown, and the only available information is a sequence of input-output couples, then learning by examples is the last chance to solve the problem. In other cases, the application might be completely defined but the complexity of a classic solution is impractical; in such cases neural techniques can be taken into account to provide a simpler solution approximating the optimal one.