Using ant colony optimisation in learning Bayesian network equivalence classes

Abstract

Bayesian networks are a useful tool in the representation of uncertain knowledge. This paper proposes a new algorithm to learn the structure of a Bayesian network. It does this by conducting a search through the space of equivalence classes of Bayesian networks using Ant Colony Optimization (ACO). To this end, two novel extensions of traditional ACO techniques are proposed and implemented. Firstly, multiple types of moves are allowed on the ACO construction graph. Secondly, moves can be given in terms of arbitrary identifiers. The algorithm is implemented and tested. The results show that ACO performs better than a greedy search whilst searching in the space of equivalence classes.