Skip to Main Content
We deal with the usage of Bayesian optimization algorithm (BOA) and its advanced variants for solving complex NP-complete combinatorial optimization problems. We focus on the hypergraph-partitioning problem and multiprocessor scheduling problem, which belong to the class of frequently solved decomposition tasks. One of the goals is to use these problems to experimentally compare the performance of the recently proposed Mixed Bayesian optimization algorithm (MBOA) with the performance of several other evolutionary algorithms based on the estimation and sampling of probabilistic model. We also propose the utilization of prior knowledge about the structure of hypergraphs and task graphs to increase the convergence speed and the quality of solutions. The performance of knowledge based MBOA (KMBOA) algorithms on the multiprocessor scheduling problem is empirically investigated and confirmed.