A NEURAL-NETWORK APPROACH TO THE DETERMINATION OF ANISOTROPY DISTRIBUTIONS

In this paper we describe methods of applying a neural network techniq ue to the magnetic inverse problem of determining the anisotropy field 's distribution from experimental transverse susceptibility data. In p rinciple, the H-K distribution represents important experimental infor mation; however, its determination is an inverse problem of great comp lexity. Using a neural network technique, we discuss methods in which the standard deviation can be determined to an accuracy which is compa rable to those of other conventional methods and completed in substant ially less time.