PARALLEL LEARNING OF BELIEF NETWORKS

Learning belief networks from a large dataset over a large domain can be computationally expensive even with a single-link lookahead search. It has been shown that a class of probabilistic domain models cannot be learned correctly by several existing algorithms which employ a single-link lookahead search. When a multi-link lookahead search is used, the computational complexity of the learning algorithm further increases. We study how to use parallelism to speed up the learning process. A parallel algorithm for learning belief networks is proposed. Our implementation in a parallel computer demonstrates the effectiveness of the algorithm.