Jan Rems: Use of relaxed stochastic controls in reinforcement learning

Datum objave: 5. 12. 2021
Seminar za verjetnost, statistiko in finančno matematiko
četrtek
9
december
Ura:
14.15
Lokacija:
Predavalnica 2.03 na FMF, Jadranska 21, Ljubljana.

V četrtek, 9. 12. 2021 ob14:15, bo v predavalnici 2.03 na FMF, Jadranska 21, potekalo predavanje Jana Remsa (FMF) z naslovom Use of relaxed stochastic controls in reinforcement learning.

Povzetek: In this talk we investigate how relaxed stochastic controls are used for exploration in continuous time and space reinforcement learning. In a special linear-quadratic case the optimal control distribution turns out to be Gaussian with mean depending on the current state, and variance depending on exploration weight parameter. A reinforcement learning algorithm for optimal investment strategy in a simple model of the financial market with the infinite horizon is presented.

Zaenkrat je predvideno, da bo predavanje potekalo v živo, bo pa organiziran tudi prenos prek videokonferenčnega sistema ZOOM. https://uni-lj-si.zoom.us/j/95492285750 ID: 954 9228 5750 Vljudno vabljeni!