作者: David M. Pennock
DOI:
关键词:
摘要: I present a parallel algorithm for exact probabilistic inference in Bayesian networks. For polytree networks with n variables, the worstcase time complexity is O(logn) on CREW PRAM (concurrent-read, exclusive-write random-access machine) processors, any constant number of evidence variables. arbitrary networks, O(r3w log n) or O(w r3w where r maximum range variable, and w induced width (the clique size), after moralizing triangulating network.