zbMATH — the first resource for mathematics

A constrained simulated annealing method for Bayesian network structure learning. (Chinese. English summary) Zbl 1289.68098
Summary: Simulated annealing method for learning Bayesian network structure is an intelligent optimization method based on the principle of scoring functions. This paper presents a constrained simulated annealing method (CSAB) to learn the Bayesian network structure. In this method, an undirected graph is obtained firstly by solving an unconstrained optimization problem; then, a simulated annealing method is used to determine the edge and its direction. Compared with the methods that use the simulated annealing method to learn Bayesian network structure directly, as the search space is reduced, this method is more efficiency.

68T05 Learning and adaptive systems in artificial intelligence
68T20 Problem solving in the context of artificial intelligence (heuristics, search strategies, etc.)
90C59 Approximation methods and heuristics in mathematical programming