A DATA FLOW IMPLEMENTATION OF AGENT-BASED DISTRIBUTED GRAPH SEARCH

Published

November 4, 2009

Author(s)

Imad Hamchi, Mathieu Hoarau, Antoine Fillinger, Nicolas Crouzier, Lukas Diduch, Martial Michel, Vincent M. Stanford

Abstract

Biological ants organize themselves into forager groups that converge to shortest paths to and from food sources. This has motivated development a large class of biologically inspired agent-based graph search techniques, called Ant Colony Optimization, to solve diverse combinatorial problems. Our approach to parallel graph search uses multiple ant agent populations distributed across processors and clustered computers to solve large-scale graph search problems. We discuss our implementation using the NIST Data Flow System II, and show good scalability of our parallel search algorithm.

Proceedings Title

Proceedings of the 2009 IASTED International Conference on Parallel and Distributed Computing Systems

Conference Dates

November 2-4, 2009

Conference Location

Cambridge , MA, US

Conference Title

IASTED International Conference on Parallel and Distributed Computing Systems

Pub Type

Conferences

Download Paper

Local Download

Keywords

NIST Data Flow System, Ant Colony Optimization (ACO), Parallel Distributed Processing, combinatorial optimization, parallel graph search.

Citation

Hamchi, I. , Hoarau, M. , Fillinger, A. , Crouzier, N. , Diduch, L. , Michel, M. and Stanford, V. (2009), A DATA FLOW IMPLEMENTATION OF AGENT-BASED DISTRIBUTED GRAPH SEARCH, Proceedings of the 2009 IASTED International Conference on Parallel and Distributed Computing Systems, Cambridge , MA, US, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=903955 (Accessed December 21, 2024)