All-Moscow scientific seminar "Mathematical Methods of Decision Analysis in Economics, Business and Politics".

On April 20 (Wednesday), 2022, a regular meeting of the all-Moscow scientific seminar "Mathematical Methods of Decision Analysis in Economics, Business and Politics" took place at the National Research University "Higher School of Economics".

Seminar leaders:
Doctor of Technical Sciences, prof. Fuad Tagievich Aleskerov,
Doctor of Technical Sciences, prof. Podinovsky Vladislav Vladimirovich,
Doctor of Technical Sciences, prof. Mirkin Boris Grigorievich.

Theme: Remarks on new otimization techniques for reinforcement learning
Speaker: Vadim Azhmyakov (HSE MIEM)
Annotation:
Our talk is devoted to the control theoretical foundations and practical implementations of advanced computational optimization methods from the Optimal Control Theory (OCT) applied to Reinforcement Learning (RL). Recently some researchers (mostly from MIT, USA) have detected a deep interconnection between the classical OCT and the conceptual part of the RL. It is common knowledge, that the conventional OCT includes many effective, rigorous, and practically approved systems optimization algorithms. This fact is due to a relatively large “uptime” of this specific control engineering area. Our main idea consists in the natural methodology transfer from the numerical OCT to the modern RL and in general to Machine Learning (ML).