TF2: A Tutorial on Optimal Algorithms for Learning Bayesian Networks
Tutorialists: James Cussens, Brandon Malone, Changhe Yuan
Monday, August 5th, afternoon
Early research on learning Bayesian networks (BNs) mainly focused on developing approximation methods such as greedy hill climbing, as the problem is shown to be NP-hard. In recent years, several exact algorithms have been developed that can learn optimal BNs for complete datasets with dozens of discrete variables. This tutorial will provide a comprehensive overview of the theoretical foundations and algorithmic developments in this important research topic, especially the recent exact algorithms. This tutorial will be of great help to students and researchers who wish to either pursue research in probabilistic graphical models or apply these methods to real-world problems.