TU Berlin

Fachgebiet Datenbanksysteme und InformationsmanagementTalk by Prof. Dr. Alexander Smola

Logo FG DIMA-new  65px


zur Navigation

“Shift your Expectations – Covariate Shift in Reinforcement Learning“ – Ein Vortrag von Prof. Dr. Alexander Smola, Direktor at Amazon Web Services, USA

Am 20. Januar hielt Prof. Dr. Alexander Smola, Direktor bei Amazon Web Services, einen Vortrag mit dem Titel „Shift your Expectations – Covariate Shift in Reinforcement Learning“.



In this talk I will give an overview over covariate shift, its correction and how it applies to reinforcement learning algorithms. RL is particularly sensitive to this phenomenon since a change in distribution is directly the result of an updated policy. That is, any update step in RL immediately results in a shift and it requires careful correcting. I will give an example of how this can lead to significant improvement in the performance of policy gradient algorithms, simply by making better use of the off-policy replay buffer. A second application illustrates how this applies to meta-learning, yielding a principled alternative to MetaRL heuristics. This is joint work with Rasool Fakoor, Pratik Chaudhari, and Stefano Soatto.


Zur Person:

Prof. Dr. Alexander Smola ist digitaler Pionier auf dem Feld des Maschinellen Lernens und Professor an der Carnegie Mellon University. Er erwarb seinen Masterabschluss an der Technischen Universität München und seinen Ph.D. an der Tu Berlin.





Schnellnavigation zur Seite über Nummerneingabe