Abductive inference in Bayesian belief networks using swarm intelligence
Karthik Ganesan Pillai, John W. Sheppard · 2012
Abductive inference in Bayesian belief networks, also known as most probable explanation (MPE) or finding the maximum a posterior instantiation (MAP), is the task of finding the most likely joint assignment to all of the (non-evidence) variables in the network. In this paper, a novel swarm intelligence-based algorithm is introduced that efficiently finds the k MPEs of a Bayesian network. Our swarm-based algorithm is compared with two state-of-the-art genetic algorithms, and the results show that the swarm-based algorithm is effective and outperforms the two genetic algorithms in terms of computational resources required.