A Novel Bayesian Network Structure Learning Algorithm: Best Parents-Children
Emna Benmohamed, Hela Ltifi, Mounir Ben Ayed · 2019
Bayesian network structure-learning represents a principal research technique in the domain of knowledge discovery. However, the construct of the correct structure is considered as an NP-hard problem due to the amount of nominee graphs, which is exponentially-related to the variables' number. To overcome this challenge, several methods had been proposed like the score and search method. In this paper, we provide an improvement of the K2 algorithm based on Best parents and children research (IK2-BN). Compared to the other algorithms in the field, and based on the obtained results, our algorithm has a better performance in terms of correct structure detection.