On Approximating K-MPE of Bayesian Networks Using Genetic Algorithm

Abstract

This paper presents a new genetic algorithm that efficiently finds k-MPE of Bayesian networks. The algorithm is based on niching method and is designed to utilize multifractral characteristic and clustering property of Bayesian networks to improve a search toward solutions. Benchmark tests are performed to evaluate the effectiveness of the algorithm and compare its performance with other niching genetic algorithms. The results from the tests show that the new algorithm outperforms the others for both running time and accuracy