Abstract

The purpose of this work is to compare different methods for solving high-dimensional linear regression problems. It is proposed to investigate various methods of accelerated and adaptive gradient descent, the conjugate gradient method, SGD and Mini-Batch SGD. The computational experiment is carried out on the sample presented in this paper. The dataset includes fMRI recordings of 30 participants aged 7 to 47 years old, obtained while they were watching a short audiovisual film.