ANALYSIS OF THE OPERATION OF DEEP NEURAL NETWORKS BY THE FAST FOURIER TRANSFORM METHOD USING C# SOFTWARE

Abstract

This article discusses the use of Fourier transform methods to analyze the performance of deep neural networks using C# software. Deep neural networks have become one of the most important tools in the field of machine learning, but their internal processes remain difficult to understand and analyze. The Fourier transform is a powerful mathematical tool that allows you to analyze signals in the frequency domain, which makes it possible to detect hidden periodic components and structures in the signals generated by neurons. The article reflects the theoretical foundations of the Fourier transform, its application for analyzing signals in deep neural networks, and also discusses in detail the software implementation of these methods in the C# programming language. In particular, numerical computing libraries such as MathNet.Numerics are used to perform an efficient discrete Fourier transform (DFT). The presented experimental results show how the Fourier transform can be used to analyze the outputs of neural networks at different stages of training, identify frequency characteristics and optimize the network architecture. Based on the conducted research, conclusions are drawn about the effectiveness of using Fourier transform methods to improve understanding of the operation of deep neural networks and their optimization.