Title :
Partitioning in reinforcement learning
Author :
Sun, Ron ; Peterson, Todd
Author_Institution :
NEC Res. Inst., Princeton, NJ, USA
Abstract :
This paper addresses automatic partitioning in complex reinforcement learning tasks with multiple agents, without a priori domain knowledge regarding task structures. Partitioning a state/input space into multiple regions helps to exploit differential characteristics of regions and differential characteristics of agents, thus facilitating learning and reducing the complexity of agents especially when function approximators are used. We develop a method for optimizing the partitioning of the space through experience without the use of a priori domain knowledge. The method is experimentally tested and compared to a number of other algorithms. As expected, we found that the multi-agent method with automatic partitioning outperformed single-agent learning
Keywords :
function approximation; learning (artificial intelligence); multi-agent systems; automatic space partitioning; function approximation; multiple agent systems; reinforcement learning; Costs; Function approximation; Jacobian matrices; Learning; National electric code; Optimization methods; Partitioning algorithms; Polynomials; State estimation; Testing;
Conference_Titel :
Neural Networks, 1999. IJCNN '99. International Joint Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-5529-6
DOI :
10.1109/IJCNN.1999.833414