A comparatively research in incremental learning of Bayesian networks
Hao Huang, Hantao Song, Fengzhan Tian, Yuchang Lu, Quande Wang · 2004
According to the way that data is processed, the learning algorithms may be classified as batch or incremental method. It is an obvious need for improving the performance and accuracy of a Bayesian network as new data is observed. While incremental learning parameters for a fixed structure have been accomplished, incremental update of Bayesian network structure is still an open problem. We have investigated the three main algorithms in incremental learning of Bayesian networks and present our theoretical analysis results. We have pointed out the main differences among the three incremental learning algorithms. Then we present our experiment result to support our theoretical analysis.