ATAL
2005
Springer
13 years 10 months ago
2005
Springer
Temporal difference (TD) learning methods [22] have become popular reinforcement learning techniques in recent years. TD methods have had some experimental successes and have been...
ISDA
2009
IEEE
13 years 11 months ago
2009
IEEE
This paper presents postponed updates, a new strategy for TD methods that can improve sample efficiency without incurring the computational and space requirements of model-based ...