S. Haider (Pakistan)
Bayesian Networks, Evolutionary Algorithms, Combinatorial Optimization, Influence Nets, Soft Computing, Computational Intelligence
The paper presents two schemes to identify the best
combination of events that maximize (minimize) the
probability of a target node in a Bayesian network (BN).
One scheme is based on evolutionary algorithms (EA)
while the other is a heuristic approach, named Sets of
Actions Finder (SAF). The heuristic approach is similar to
the hill-climbing search technique and works in
polynomial time. Both SAF and EA are applied over
hundreds of Bayesian networks and the results are
compared with the ones obtained through exhaustive
searches. The results show that both SAF and EA perform
quite similarly but the time taken by SAF is smaller
compared to EA.