| Abstract |
The large-scale matter distributions tell us the evolution of the Universe as well as its components and initial state. Deep learning can be useful in analyzing observational data of large-scale structures. We have been developing new deep-learning methods for (i) removing noises in observed data and (ii) inferring the governing parameters from observed data. In this talk, I will introduce these methods and also discuss how to tackle the problem of large data requirements in deep learning. |
(20.77MB)