Classification with Belief Decision Trees

Decision trees are considered as an efficient technique to express classification knowledge and to use it. However, their most standard algorithms do not deal with uncertainty, especially the cognitive one. In this paper, we develop a method to adapt the decision tree technique to the case where the object’s classes are not exactly known, and where the uncertainty about the class’ value is represented by a belief function. The adaptation concerns both the construction of the tree and its use to classify new objects characterized by uncertain attribute values.