On a statistical learning algorithm of superstatistics. Issue 3 (22nd January 2020)

• Record Type: Journal Article
• Authors: Alzeley, Omar
• Is Part Of: Computational and mathematical methods. Volume 2:Issue 3(2020)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-01-22
• Subjects: characteristic function -- neural network -- superstatistics; Mathematics -- Data processing -- Periodicals; Numerical analysis -- Periodicals; Numerical analysis; Mathematics -- Data processing; Periodicals; 004.0151
• DOI: 10.1002/cmm4.1085
• Languages: English
• ISSNs: 2577-7408

Abstract

Abstract : The purpose of this article is to investigate the ability of a neural network to deal with a statistics of statistics. The idea behind our approach is that the neural network should become capable of dealing with this kind of method through the use of a learning process. The novel algorithm used herein is based on a characteristic function related to a Fourier transform. However, we argue that this characteristic function is more useful than the Laplace transform since its finiteness is guaranteed. We demonstrate the veracity of this algorithm by applying it to some well‐known distributions. Our algorithm is capable of handling both continuous and discrete distributions, and through which the neural network is able to learn and predict the distribution of interest.