Efficient Bayesian Network Inference: Genetic Algorithms, Stochastic Local Search, and Abstraction

210 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1999.Two major research results are presented that relate to creating hard synthetic Bayesian networks for empirical research on inference algorithms. One method translates deceptive problems studied in genetic algorithms to a Bayesian network setting, showing that Bayesian networks can be deceptive. The other result is based on translating satisfiability problems into Bayesian networks. We describe how connectivity, value of conditional probability tables as well as the degree of regularity of the underlying graph affect the speed of inference for Hugin and Stochastic Greedy Search.U of I OnlyRestricted to the U of I community idenfinitely during batch ingest of legacy ETD