Machine-learning applications for weak-lensing cosmology

TL;DR

This review discusses how machine learning enhances weak-lensing cosmology by overcoming traditional analysis limitations and improving the extraction of cosmological information from galaxy-imaging surveys.

Contribution

It summarizes recent advances in applying machine learning to weak-lensing cosmology and outlines future prospects for these techniques.

Findings

Machine learning mitigates limitations of traditional weak-lensing analysis.

ML techniques improve the extraction of cosmological information.

Future ML approaches can enhance data analysis for upcoming surveys.

Abstract

This article reviews recent advances in the application of machine learning to weak-lensing cosmology. Weak gravitational lensing provides a unique and powerful probe of the total matter distribution in the Universe, independent of its physical state. By directly tracing the spatial distribution of otherwise invisible dark matter within the cosmic web, weak lensing has become a cornerstone for studying both the nature of dark matter and the physics governing large-scale structure formation. We begin by introducing the conventional estimators used to extract weak-lensing signals from modern galaxy-imaging surveys and by summarizing established methods for deriving cosmological information from these observables. We then discuss the limitations inherent in traditional analyses and outline how machine-learning techniques can mitigate these challenges. Finally, we explore future prospects for machine-learning-based approaches, highlighting their potential to further enhance the scientific return of current and upcoming weak-lensing datasets.