Jump to Navigation

Machine Learning Approaches to Multi-Wavelength Studies of MeerKAT-MIGHTEE Radio Sources

Astrophysics is undergoing a data deluge. Over the past decade the advent of large-format digital de- tectors and automated survey telescopes have increased the size and complexity of datasets available for astrophysical studies by several orders of magnitude. Making sense of these datasets calls for a new approach toward the automated study of many millions (and soon billions) of astronomical objects, and Machine Learning algorithms that can ’learn’ from the data with a small degree (if any) of interac- tion/supervision on the scientist’s side are increasingly being used. Extragalactic radio surveys currently getting started with MeerKAT, the South African precursor to the Square Kilometre Array (SKA), will provide unprecedented radio continuum, spectral line and polariza- tion information for us to study the formation and evolution of Galaxies and AGNs in the distant Universe. The rapid growth of radio data from MeerKAT surveys thus provides an excellent opportunity to apply Machine Learning approaches. In this MSc project, we will optimise machine learning techniques to the identification and characterization of MeerKAT radio sources detected in the MeerKAT International GHz Tiered Extragalactic Exploration (MIGHTEE) survey. MIGHTEE (https://idia.ac.za/mightee/) is a project being undertaken by an SA-led international collaboration of researchers to explore Galaxy and AGN evolution over cosmic time with MeerKAT and covering over 20 deg2 of well-studied extragalactic fields. The student will optimise our machine learning methods (An et al. 2018) to identify multi- wavelength counterparts of MIGHTEE radio sources and study their physical properties. The project is expected to lead to the development of original machine learning software as well as to a publication. This project will optimize machine learning algorithms for use with MeerKAT data some of the largest multi-wavelength extragalactic survey datasets obtained to date to answer key science questions in modern galaxy evolution studies. It will thus allow the student to develop a solid background in ma- chine learning and in multi-wavelength astronomy, but it will also have important implications for sci- ence topics to be pursued with MeerKAT and SKA survey projects, and it would thus lend itself to be upgraded to a PhD project. The student will be co-supervised by Associate Professor and eRe- search Director Mattia Vaccari and Postdoctoral Research Fellow Dr Fangxia An within the HIPPO (http://www.mattiavaccari.net/hippo/) research group at UWC where (s)he will have access to IDIA facilities (http://www.idia.ac.za). The project requires a good understanding of extragalactic astron- omy and a good proficiency in python software development as well as the willingness to develop both. Please get in touch over e-mail to discuss the project in person! References : Djorgovski et al. 2013, ”Sky Surveys”, http://arxiv.org/pdf/1203.5111v2.pdf; Ball & Brunner 2010, ”Data Mining and Machine Learning in Astronomy”, http://arxiv.org/pdf/0906. 2173.pdf; Jarvis, Taylor et al. 2016, ”The MeerKAT International GHz Tiered Extragalactic Exploration (MIGHTEE) Survey”, https://pos.sissa.it/277/006/pdf, An et al. 2018, ”A Machine-learning Method for Identifying Multiwavelength Counterparts of Submillimeter Galaxies”, https://arxiv.org/ pdf/1806.06859.pdf; Liu et al. 2019, ”A machine learning approach for identifying the counterparts of submillimetre galaxies and applications to the GOODS-North field”, https://arxiv.org/pdf/1901. 09594.pdf; scikit-learn tutorials, https://scikit-learn.org/stable/tutorial/index.html.
Node This Project Is Offered On: 
Project Info: 
PDF icon Project Description




Research Area: 

Available_projects | by Dr. Radut