\u3010Date\u3011
10:00-11:30, Wed. Oct. 19<\/p>\n
\u3010Speaker\u3011
Kana MORIWAKI \/The University of Tokyo<\/p>\n
\u3010Title\u3011
\"Deep Learning Application for Reconstruction of Large-Scale Structure of the Universe\"<\/p>\n
\u3010Abstract\u3011
Deep learning has been widely used in various research fields including astrophysics and cosmology. In this talk, I will introduce a deep learning approach for reconstructing the distributions of galaxies and matter over a cosmological volume from observational data. Such distributions, called large-scale structures (LSSs) of the universe, provide information on the evolution of the universe. We develop a generative model based on three-dimensional convolutional neural network to extract the LSS signals from noisy observational data. By adopting physics-informed<\/i> network architecture, the information extraction is performed more accurately. The talk will also address the challenges of applying machine learning methods to future astronomical observation data for which the governing physical laws are not yet fully understood.<\/p>\n