The development of machine learning methods and their success in a wide range of problems have had a significant impact on the design and implementation of solar physics research. Large data sets have emerged as an intrinsic value in which the efforts of experts and significant technological resources are invested. The research itself has acquired an interdisciplinary nature and is concentrated around advanced computing centers. Large-scale problems can now be posed whose mathematical formulation was unclear yesterday. In this review, we present the main ideas underlying modern machine learning models, the databases prepared for machine learning tasks, and data processing tools. A major part of this review is devoted to a discussion of models proposed in the context of specific solar physics problems and their extension to other applications.

Typically, an English full text is available in about 1 month from the date of publication of the original article.

Correspondence should be addressed to † egor.illarionov@math.msu.ru Keywords: solar physics, solar activity, machine learning, databases PACS: 07.05.Mh, 84.35.+i, 96.60.−j (all) DOI: 10.3367/UFNe.2025.02.039872 URL: https://ufn.ru/en/articles/2025/4/e/ Citation: Illarionov E A "Machine learning methods in solar physics" Phys. Usp. 68 (4) (2025) Received: 5th, August 2024, accepted: 22nd, February 2025 Оригинал: Илларионов Е А «Методы машинного обучения в физике Солнца» УФН 195 395–415 (2025); DOI: 10.3367/UFNr.2025.02.039872