Multi-Variable Agent decomposition for DCOPs
Ferdinando Fioretto, William Yeoh, Enrico Pontelli · 2016
The application of DCOP models to large problems faces two main limitations: (i) Modeling limitations, as each agent can handle only a single variable of the problem; and (ii) Res-olution limitations, as current approaches do not exploit the local problem structure within each agent. This paper pro-poses a novel Multi-Variable Agent (MVA) DCOP decompo-sition technique, which: (i) Exploits the co-locality of each agent’s variables, allowing us to adopt efficient centralized techniques within each agent; (ii) Enables the use of hier-archical parallel models and proposes the use of GPUs; and (iii) Reduces the amount of computation and communication required in several classes of DCOP algorithms.