SOURCE-LINKED INTELLIGENCE

Mapping the Extreme Universe with deep neural networks: from simulations to Rubin-LSST data

CORDIS · observation · Publication date unknown

and broadening the Hungarian contributions to the next-generation Legacy Survey of Space and Time (LSST) project based in Chile. To go beyond the state-of-the-art, I will acquire extensive skills on machine learning techniques from expert researchers at Konkoly Observatory to combine with my groundwork results on cosmological data analysis from, above all, the Dark Energy Survey (DES). As a key innovation, I will develop deep learning models to study extreme voids and superclusters. First, I will apply convolutional neural network methods to augment traditional cross-correlations between galaxy density fluctuations and the anisotropies of the Cosmic Microwave Background. Then, I will capture the dependence of their gravitational signals on the physical properties of dark energy and dark matter. The proposed analyses of simulations and early observational LSST data will help resolve whether some as-yet unknown physical effects or systematic biases complicate the picture in cosmology. Either way we will gather fundamentally new knowledge about the Universe on the largest scales. cosm

Read original source ↗ Open in workspace

recordType

award

status

SIGNED

region

EU

value

141782.4

unit

EUR