Harnessing data using symbolic regression methods for discovering novel paradigms in physics

In recent years, machine-learning methods have profoundly impacted research in the interdisciplinary fields of physics. However, most machine-learning models lack interpretability, and physicists doubt the credibility of their conclusions because they cannot be combined with prior physical knowledge. Therefore, this review focuses on symbolic regression, which is an interpretable machine-learning method. First, the relevant concepts of machine learning are introduced in conjunction with induction. Next, we provide an overview of symbolic regression methods. Subsequently, the recent directions for the application of symbolic regression methods in different subfields of physics are outlined, and an overview of the ways in which the applications of symbolic regression have evolved in the realm of physics is provided. The major aim of this review is to introduce the basic principles of symbolic regression and explain its applications in the field of physics.