Decision support in image mining for vague objects

A critical issue in image mining concerns communication to users, e.g. by decision making. In this paper, we address decision making on vague objects. We first present image mining for uncertain (vague) objects. Uncertain objects are often modeled as fuzzy sets, requiring definition, estimation and use of membership functions. In this study, a Bayesian approach is presented, in which a prior estimate of the linear parts of the membership function is adjusted using observed data. Thus, a more solid way is found to use fuzzy and vague objects in decision support, i.e. in communication to users.