A New Pareto-Based Algorithm for Multi-objective Graph Partitioning

摘要

One significant problem of optimization which occurs in many real applications is that of graph partitioning. It consist of obtaining a partition of the vertices of a graph into a given number of roughly equal parts, whilst ensuring that the number of edges connecting vertices of different sub-graphs is minimized. In the single-objective (traditional) graph partitioning model the imbalance is considered a constraint. However, in same applications it is necessary to extend this model to its multi-objective formulation, where the imbalance is also an objective to minimize. This paper try to solve this problem in the multi-objective way by using a population version of the SMOSA algorithm in combination with a diversity preservation method proposed in the SPEA2 algorithm.