A local cost simulation-based algorithm to solve distributed constraint optimization problems

As an important incomplete algorithm for solving Distributed Constraint Optimization Problems (DCOPs), local search algorithms exhibit the advantages of flexibility, high efficiency and high fault tolerance. However, the significant historical values of agents that affect the local cost and global c...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:PeerJ Computer Science Ročník 9; s. e1296
Hlavní autoři: Shi, Meifeng, Liang, Feipeng, Chen, Yuan, He, Ying
Médium: Journal Article
Jazyk:angličtina
Vydáno: United States PeerJ 17.03.2023
PeerJ. Ltd
PeerJ Inc
Témata:
Agents and Multi-Agent Systems
Algorithms
Analysis
DCOPs
Economic aspects
Electronic computers. Computer science
EWMA
Inferior solutions
LCS
Optimization Theory and Computation
QA75.5-76.95
Inferior solutions
LCS
EWMA
DCOPs
ISSN:2376-5992, 2376-5992
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:As an important incomplete algorithm for solving Distributed Constraint Optimization Problems (DCOPs), local search algorithms exhibit the advantages of flexibility, high efficiency and high fault tolerance. However, the significant historical values of agents that affect the local cost and global cost are never taken into in existing incomplete algorithms. In this article, a novel Local Cost Simulation-based Algorithm named LCS is presented to exploit the potential of historical values of agents to further enhance the exploration ability of the local search algorithm. In LCS, the Exponential Weighted Moving Average (EWMA) is introduced to simulate the local cost to generate the selection probability of each value. Moreover, populations are constructed for each agent to increase the times of being selected inferior solutions by population optimization and information exchange between populations. We theoretically analyze the feasibility of EWMA and the availability of solution quality improvement. In addition, based on our extensive empirical evaluations, we experimentally demonstrate that LCS outperforms state-of-the-art DCOP incomplete algorithms.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:2376-5992
2376-5992
DOI:10.7717/peerj-cs.1296