In the presentation, in the first part, we will talk about some numerical methods in order to solve non-smooth convex optimization problems with functional constrained in deterministic, stochastic, and online settings. In order to solve such classes of problems, it was proposed a new modification of Adaptive Mirror Descent algorithms in both deterministic and stochastic settings. It was obtained specific estimates of the convergence rate of some Mirror Descent algorithms for different levels of smoothness of objective function.
In the second part of the presentation, we will talk about gradient-type adaptive methods for relatively Lipschits and relatively smooth convex optimization problems.

Link to the seminar is: 

https://teams.microsoft.com/l/meetup-join/19%3a0ae91f7b86a24e8fa1d818a6f74b22a1%40thread.tacv2/1634560208841?context=%7b%22Tid%22%3a%22066690f2-a8a6-4889-852e-124371dcbd6f%22%2c%22Oid%22%3a%22a343f6fd-86f8-4abe-95cf-3c7f4ad5f0ca%22%7d