Deep Machine Learning in Cosmology: Evolution or Revolution?
Ofer Lahav
备注：
12 pages, 11 figures, Plenary lecture at the IAU GA (August 2022, South Korea), to appear in Proceedings of IAU Symposium 368, 'Machine Learning in Astronomy: Possibilities and Pitfalls'
Could Machine Learning (ML) make fundamental discoveries and tackle unsolved problems in Cosmology? Detailed observations of the present contents of the universe are consistent with the Cosmological Constant Lambda and Cold Dark Matter model, subject to some unresolved inconsistencies ('tensions') among observations of the Hubble Constant and the clumpiness factor. To understand these issues further, large surveys of billions of galaxies and other probes require new statistical approaches. In recent years the power of ML, and in particular 'Deep Learning', has been demonstrated for object classification, photometric redshifts, anomaly detection, enhanced simulations, and inference of cosmological parameters. It is argued that the more traditional 'shallow learning' (i.e. with pre-processing feature extraction) is actually quite deep, as it brings in human knowledge, while 'deep learning' might be perceived as a black box, unless supplemented by explainability tools. The 'killer applications' of ML for Cosmology are still to come. New ways to train the next generation of scientists for the Data Intensive Science challenges ahead are also discussed. Finally, the chatbot ChatGPT is challenged to address the question posed in this article's title.
Harvard University（哈佛大学）；The NSF AI Institute for Artificial Intelligence and Fundamental Interactions（美国国家科学基金会人工智能与基础交互研究所）；Massachusetts Institute of Technology（麻省理工学院）；Fermi National Accelerator Laboratory（费米国家加速器实验室）；University of Chicago（芝加哥大学）；The Pennsylvania State University（宾夕法尼亚州立大学）；University of Michigan（密歇根大学）；University of Wisconsin-Madison（威斯康星大学麦迪逊分校）；University of Washington（华盛顿大学）
Machine Learning Closure Audits for LSST Photometric Supernova Cosmology
Center for AstroPhysical Surveys, National Center for Supercomputing Applications（国家超级计算应用中心天体物理调查中心）；Department of Astronomy, University of Illinois at Urbana Champaign（伊利诺伊大学厄巴纳-香槟分校天文系）
Cosmology with galaxy clusters using machine learning. Application to eROSITA Data
University Federico II（那不勒斯费德里科二世大学）；School of Physics and Astronomy, Sun Yat-sen University（中山大学物理与天文学院）；INAF – Osservatorio Astronomico di Capodimonte（意大利国家天体物理研究所卡波迪蒙特天文台）；Scuola Superiore Meridionale（南方高等学院）；Univ. Grenoble Alpes（格勒诺布尔阿尔卑斯大学）；CNRS（法国国家科学研究中心）；Grenoble INP（格勒诺布尔国立理工学院）；LPSC-IN2P3（粒子与核物理实验室）；Max-Planck-Institut für Astrophysik（马克斯·普朗克天体物理学研究所）；Universitäts-Sternwarte, Fakultät für Physik, Ludwig-Maximilians-Universität München（慕尼黑大学物理学院大学天文台）；Institut für Theoretische Physik, Universität Heidelberg（海德堡大学理论物理研究所）
A Systematic Literature Review of Machine Learning Techniques for Observational Constraints in Cosmology
Universidad San Sebastián（圣塞巴斯蒂安大学）；Universidad del Bío-Bío（比奥比奥大学）；Universidad Católica del Norte（北天主教学会大学）；East China University of Science and Technology（华东理工大学）
Towards a Robust Machine-Learning Pipeline for 21-cm Cosmology Data Analysis I: A Roadmap for Development and Demonstration of Robustness Against PSF Modeling Errors
Machine Learning for Multi-messenger Probes of New Physics and Cosmology: A Review and Perspective
Center for Theoretical Physics, College of Physics Science and Technology, Sichuan University（四川大学物理科学与技术学院理论物理中心）；INFN, Laboratori Nazionali di Frascati（意大利国家核物理研究所弗拉斯卡蒂国家实验室）；National Research Nuclear University “MEPHI”（莫斯科国立核研究大学（MEPHI））；Virtual Institute of Astroparticle Physics（虚拟高能天体物理研究所）；Research Institute of Physics, Southern Federal University（南联邦大学物理研究所）；College of Design and Innovation, Tongji University（同济大学设计创意学院）；Department of Physics, University of Rome “La Sapienza”（罗马第一大学物理系）；Istituto Nazionale Fisica Nucleare, Sezione di Roma（意大利国家核物理研究所罗马分部）；College of Arts, Media, and Technology, Chiang Mai University（清迈大学艺术、媒体与技术学院）
Cosmology with Persistent Homology: Parameter Inference via Machine Learning
Department of Physics, University of Wisconsin-Madison（威斯康星大学麦迪逊分校物理系）；Center for Astrophysics and Cosmology, University of Nova Gorica（诺瓦戈里察大学天体物理学与宇宙学中心）
