Issue No. 02 - February (1996 vol. 45)

ISSN: 0018-9340

pp: 195-208

DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/12.485372

ABSTRACT

<p><b>Abstract</b>—Given a graph <b>G</b>, we intend to partition its nodes into two sets of equal size so as to minimize the sum of the cost of the edges having end-points in different sets. This problem, called the uniform graph partitioning problem, is known to be NP-Complete. In this paper we propose the first reported learning-automaton based solution to the problem. We compare this new solution to various reported schemes such as the Kernighan-Lin's algorithm, and two excellent recent heuristic methods proposed by Rolland et al.—an extended local search algorithm and a genetic algorithm. The current automaton-based algorithm outperforms all the other schemes. We believe that it is the fastest algorithm reported to date. Additionally, our solution can also be adapted for the GPP (See note at end of Section 1) in which the edge costs are not <it>constant</it> but random variables whose distributions are <it>unknown</it>.</p>

INDEX TERMS

Heuristic search; graph partitioning; adaptive learning; learning automata.

CITATION

Edward V. de St. Croix, B. John Oommen, "Graph Partitioning Using Learning Automata",

*IEEE Transactions on Computers*, vol. 45, no. , pp. 195-208, February 1996, doi:10.1109/12.485372