Objectives:
Statistical learning theory is mathematical foundation of machine learning - the currently fastest growing branch of computer sciences and artificial intelligence. Central objects of statistical learning theory are statistical models. The project is based on our results obtained jointly with N. Ay and J. Jost and covers the following topics: geometry of efficient estimations, geometry of natural gradient flows and properties of Kullback-Leibler divergence on statistical models, in particular graphical models, hidden Markov models, Boltzmann machine, multilayer perceptrons and infinite dimensional exponential models.
Main investigator:
Participating institutions:
Institute of Mathematics, AS CR
Site map | Author | Webmaster
