Showing 1–1 of 1 results for author: Hoang, K D
-
New Algorithms for Functional Distributed Constraint Optimization Problems
Authors:
Khoi D. Hoang,
William Yeoh,
Makoto Yokoo,
Zinovi Rabinovich
Abstract:
The Distributed Constraint Optimization Problem (DCOP) formulation is a powerful tool to model multi-agent coordination problems that are distributed by nature. The formulation is suitable for problems where variables are discrete and constraint utilities are represented in tabular form. However, many real-world applications have variables that are continuous and tabular forms thus cannot accurate…
▽ More
The Distributed Constraint Optimization Problem (DCOP) formulation is a powerful tool to model multi-agent coordination problems that are distributed by nature. The formulation is suitable for problems where variables are discrete and constraint utilities are represented in tabular form. However, many real-world applications have variables that are continuous and tabular forms thus cannot accurately represent constraint utilities. To overcome this limitation, researchers have proposed the Functional DCOP (F-DCOP) model, which are DCOPs with continuous variables. But existing approaches usually come with some restrictions on the form of constraint utilities and are without quality guarantees. Therefore, in this paper, we (i) propose exact algorithms to solve a specific subclass of F-DCOPs; (ii) propose approximation methods with quality guarantees to solve general F-DCOPs; and (iii) empirically show that our algorithms outperform existing state-of-the-art F-DCOP algorithms on randomly generated instances when given the same communication limitations.
△ Less
Submitted 30 May, 2019;
originally announced May 2019.
