Νέα / Ανακοινώσεις / Συζητήσεις

Ομιλία Καθηγητή Μιχαήλ Κατεχάκη (Rutgers University, USA)

  • 1
  • Συντάχθηκε 02-10-2019 10:41 Πληροφορίες σύνταξης

    Ενημερώθηκε: -

    Τίτλος

    Reinforcement Learning: The state of the Art 

    Ομιλητής

    Michael N. Katehakis
    Distinguished Professor and Chair
    Department of Management Science and Information Systems
    Rutgers Business School Newark and New Brunswick

    Περίληψη

    Reinforcement Learning (RL) refers to  techniques designed for sequential decision making when a system needs to "learn" a strategy  that maximizes  a  reward  (or minimizes  a cost) criterion when some parameters of the basic underlying  model are not known in advance. RL    is experiencing significant growth in recognition due to successful applications in many areas of machine learning (ML).

    In this talk we provide a  survey of the state of the art of the area of computing optimal data driven RL algorithms. Then, we compare the performance of the classic UCB policy of Burnetas and Katehakis (1987) new algorithms recently proposed: optimistic programming, the MDP-Deterministic Minimum Empirical Divergence (MDP-DMED), as well as a method based on Posterior sampling (MDP-PS).  

    We also discuss the origins of RL and its connection with model and theory of the so-called multi-armed bandit problem.


    Τόπος: Λ - Κτίριο Επιστημών/ΗΜΜΥ, 137Π-39,-38, Ισόγειο Πυρήνα
    Έναρξη: 02/10/2019 14:00
    Λήξη: 02/10/2019 15:00


  • 1

Σύνδεση χρήστη

Εισάγετε εδώ το όνομα χρήστη και τον κωδικό πρόσβασής σας για να συνδεθείτε στον ιστότοπο.
Σύνδεση
© Πολυτεχνείο Κρήτης 2012