LEARNING MEMBERSHIP FUNCTIONS IN A FUNCTION-BASED OBJECT RECOGNITION SYSTEM

Functionality-based recognition systems recognize objects at the categ ory level by reasoning about how well the objects support the expected function. Such systems naturally associate a ''measure of goodness'' or ''membership value'' with a recognized object. This measure of good ness is the result of combining individual measures, or membership val ues, from potentially many primitive evaluations of different properti es of the object's shape. A membership function is used to compute the membership value when evaluating a primitive of a particular physical property of an object. In previous versions of a recognition system k nown as GRUFF, the membership function for each of the primitive evalu ations was hand-crafted by the system designer. In this paper, we prov ide a learning component for the GRUFF system, called OMLET, that auto matically learns membership functions given a set of example objects l abeled with their desired category measure. The learning algorithm is generally applicable to any problem in which low-level membership valu es are combined through an aad-or tree structure to give a final overa ll membership value.