Improvement for TANC-BIC Structure Learning Algorithm
Hao Xu · Computer Technology and Development · 2006
Bayesian classifier based on probability theory has gained great attention,because of its simple structure and good performance.TANC applies widely in practice.The classifier which was set up by the TANC-BIC structure-learning algorithm had acquired success,but it didn't consider the class node.This paper suggests a new structure-learning algorithm called TANC-CBIC,makes experiment in MBNC experiment platform with programming TANC-CBIC algorithm.The results show that the accuracy of improver is better than algorithm based onTANC-BIC and TANC-CMI.The new structure-learning algorithm is effective.