Method for modelling synthetic data in generative adversarial networks

Abstract

A method for modelling a shape of data in a GAN comprising a generator agent (G) for generating synthetic data () and a discriminator agent (D) for distinguishing between the generated synthetic data () and real original data () that follow an arbitrary, continuous or discrete, distribution defined by a n-dimensional vector of input variables x. The method, before generating output synthetic data (), computes an Inverse Smirnov transformation ffor each of the n input variables xand attaches an activation function () to the generator agent (G), wherein the activation function () is a n-dimensional vector fformed by the computed Inverse Smirnov transformations, f=(f, f, . . . , f). The GAN implementing the method generates the output synthetic data () using the activation function () which outputs synthetic data () with a distribution (') whose shape is the same as the arbitraty distribution of the original data ().