97 search results - page 5 / 20 » An epsilon-Optimal Grid-Based Algorithm for Partially Observ... |
COLT
2000
Springer
13 years 11 months ago
2000
Springer
We model reinforcement learning as the problem of learning to control a Partially Observable Markov Decision Process ( ¢¡¤£¦¥§ ), and focus on gradient ascent approache...
ATAL
2010
Springer
13 years 8 months ago
2010
Springer
In this paper, we study a particular subclass of partially observable models, called quasi-deterministic partially observable Markov decision processes (QDET-POMDPs), characterize...
IJCAI
2003
13 years 8 months ago
2003
We address the problem of optimally controlling stochastic environments that are partially observable. The standard method for tackling such problems is to define and solve a Part...
PERCOM
2007
ACM
14 years 7 months ago
2007
ACM
We consider sensor scheduling as the optimal observability problem for partially observable Markov decision processes (POMDP). This model fits to the cases where a Markov process ...
FOCS
2007
IEEE
14 years 1 months ago
2007
IEEE
We consider a variant of the classic multi-armed bandit problem (MAB), which we call FEEDBACK MAB, where the reward obtained by playing each of n independent arms varies according...