Generalizing DPOP: Action-GDL, a new complete algorithm for DCOPs

In this paper we propose a novel message-passing algorithm, the so-called Action-GDL, as an extension to the Generalized Distributive Law algorithm (GDL) [1] to efficiently solve DCOPs. We show the generality of Action-GDL by proving that it has DPOP, one of the low-complexity, state-of-the-art algorithm to solve DCOPs, as a particular case. Finally, we provide empirical evidences to illustrate how Action-GDL can outperform DPOP in terms of computation, communication and parallelism needed to solve the problem.

• Type: Conference
• Year: 2009
• Where: ATAL
• Authors: Meritxell Vinyals, Juan A. Rodríguez-Aguilar, Jesús Cerquides
• Keywords: Artificial Intelligence, ATAL 2009, Distributive Law Algorithm, So-called Action-gdl, State-of-the-art Algorithm