71 search results - page 4 / 15 » An Analysis of Direct Reinforcement Learning in Non-Markovia... |
ATAL
2009
Springer
14 years 2 months ago
2009
Springer
In several agent-oriented scenarios in the real world, an autonomous agent that is situated in an unknown environment must learn through a process of trial and error to take actio...
ICML
1995
IEEE
14 years 8 months ago
1995
IEEE
A number of reinforcement learning algorithms have been developed that are guaranteed to converge to the optimal solution when used with lookup tables. It is shown, however, that ...
NAACL
2001
13 years 9 months ago
2001
Developing dialogue systems is a complex process. In particular, designing efficient dialogue management strategies is often difficult as there are no precise guidelines to develo...
ECML
2007
Springer
13 years 11 months ago
2007
Springer
Abstract. We present a new reinforcement learning approach for deterministic continuous control problems in environments with unknown, arbitrary reward functions. The difficulty of...
ICML
2005
IEEE
14 years 8 months ago
2005
IEEE
We introduce flexible algorithms that can automatically learn mappings from images to actions by interacting with their environment. They work by introducing an image classifier i...