Advanced search
Start date

Machine learning techniques applied to cosmological problems

Grant number: 19/08852-2
Support type:Scholarships in Brazil - Post-Doctorate
Effective date (Start): June 01, 2019
Effective date (End): May 31, 2021
Field of knowledge:Physical Sciences and Mathematics - Physics
Principal Investigator:Nathan Jacob Berkovits
Grantee:Martín Emilio de Los Rios
Home Institution: Instituto de Física Teórica (IFT). Universidade Estadual Paulista (UNESP). Campus de São Paulo. São Paulo , SP, Brazil


Machine learning techniques represents a new way of analysing big datasets in an agnostic and homogeneous way. These methods are very useful and powerful tolls to find patterns and relations between the variables that are involved in an specific problem. It is worth to mention that these techniques have been applied with a lot of success in technological problems and in other different areas of science, including astronomy and physics.On the other hand, current and future astronomic surveys will generate enormous amount of information, making the machine learning techniques important tools for their analysis.During this postdoc I will applied this new techniques to different cosmological problems.Specifically I will improve the machine learning algorithms for the automatic classification of merging clusters, in order to applied it for high-redshift and Sunyaev-Zeldovich clusters. At the same time I will continue studying in an individual way the merging clusters candidates found previously.I also will continue the studies of the cosmic microwave background anisotropies with the aimof find if there are any signal of the departure from the standard cosmological model. Specifically I will apply anomaly detection algorithms to find zones of the sky which may present deviations from what it is expected.