A method to induce bayesian networks from data to overcome some limitations of other learning algorithms is proposed. One of the main features of this method is a metric to evaluate bayesian networks combining different quality criteria. A fuzzy system is proposed to enable the combination of different quality metrics. In this fuzzy system a metric of classification is also proposed, a criterium that is not often used to guide the search while learning bayesian networks. Finally, the fuzzy system is integrated to a genetic algorithm, used as a search method to explore the space of possible bayesian networks, resulting in a robust and flexible learning method with performance in the range of the best learning algorithms of bayesian networks developed up to now.

No abstract

No abstract

We introduce a family of divergences between Φ-probabilistic sets, with real supports. The supports are never unbounded to opposite sides. We start from weighted and percentiled dissimilarities between arbitrary unions of compact intervals of real numbers. As an application we model the problem of the recognition of a handshape as a metric problem between Φ-probabilistic sets. The proposed family of divergences is a suitable solution to this problem of comparing one handshape prototype, a Φ-probabilistic set, with one input handshape, a Φ-fuzzy set.

No abstract