A neural network approach to the determination of anisotropy distributions

In this paper we describe methods of applying a neural network technique to the magnetic inverse problem of determining the anisotropy field's distribution from experimental transverse susceptibility data. In principle, the distribution represents important experimental information; however, its determination is an inverse problem of great complexity. Using a neural network technique, we discuss methods in which the standard deviation can be determined to an accuracy which is comparable to those of other conventional methods and completed in substantially less time.