Speaker
Boris Prokhorov
(Moscow Institute of Physics and Technology)
Description
In this work we study lower bounds for optimal convergence rates in stochastic
optimization with smooth strongly convex objective function and first-order markovian oracle. In these settings we provided lower bounds matching previously known
upper bounds for a wide family of target functionals. Optimal rates remain unknown
only for the special case of small noise.
Primary author
Boris Prokhorov
(Moscow Institute of Physics and Technology)
Co-author
Aleksandr Beznosikov
(Moscow Institute of Physics and Technology)
