PhD Seminar: Recovering hidden parameters of cyber-physical systems from observable characteristics
Restoring hidden parameters from observations is important in various fields. We will briefly talk about reconstructing the rheology of the environment, modeling the internal parameters of a data storage system, and, in more detail, about reconstructing the parameters of the true positions of satellites from observed ones by telescopes. This problem was solved for the Russian Astronomical Science Center and was presented at the HSE International Data Analysis Olympiad (IDAO) in 2020. Among the technologies used are machine learning and, in particular, deep learning.