摘要
针对SP算法中利用优化组合处理稀疏候选集来评分得最优候选集,这样得到的每个节点的候选集为父节点集,从而容易导致最后的贝叶斯网络双向边较多,对双向边处理后还存在较多的反向边,从而提出了利用爬山算法处理稀疏候选集,得到新的算法SCHC,该算法减少了双向边的数量和提高了正确边的数量。
SP algorithm used optimization to deal with sparse candidate sets, scored the optimal .set as the candidate, so are the candidates for each node - parent node sets, thus easily lead to the final Bayesian network had more two-way edges, after dealed with two - way edge, there are still more reverse side. And put forward using hill-climing algorithm to deal with sparse candidate, sets,getting new algo-rithm SCHC. And the algorithm has reduced the number of two- way edges and increased the number of correct side.
出处
《计算机技术与发展》
2009年第3期155-157,192,共4页
Computer Technology and Development
基金
国家自然科学基金(39880032)
关键词
SP算法
稀疏候选集
贝叶斯网络
爬山算法
双向边
SP algorithm
sparse candidate sets
bayesian networks
hill - climing algorithm
Two- way edges