NEURAL NETWORKS AS A TOOL FOR MATHEMATICAL COMPUTATION

Abstract

Abstract. Today, the application of neural networks has become ubiquitous almost everywhere. But not many researchers in a field far from mathematics notice that there is a huge mathematical background behind these deep neural networks. This shortcoming causes some unintended difficulties during the research. This is because the fundamental basis of each recorded neural network model ranges from simple mathematical expressions to complex levels. We show how these mathematical aspects are applied and play an important role in neural network models, revealing their application to educators in general, experts in each field, to at least somewhat fill these gaps in this paper.