Classification trees based on belief functions

Nicolas Sutton Charani (1) , Sébastien Destercke (1) , Thierry Denoeux (1)

1 Heudiasyc - Heuristique et Diagnostic des Systèmes Complexes [Compiègne]

Decision trees classifiers are popular classification methods. In this paper, we extend to multi-class problems a decision tree method based on belief functions previously described for 2-class problems only. We propose two ways to achieve this extension: combining multiple 2-class trees together and directly extending the estimation of belief functions within the tree to the multi-class setting. We provide experiment results and compare them to classical decision trees.

Mots clés: Belief functions; Classification; Decision trees

Domaines: Apprentissage [cs.LG]; Intelligence artificielle [cs.AI]; Machine Learning [stat.ML]; Statistiques [math.ST]; Théorie [stat.TH]

Nicolas Sutton Charani, Sébastien Destercke, Thierry Denoeux. Classification trees based on belief functions. 2nd International Conference on Belief Functions (BELIEF 2012), May 2012, Compiègne, France. pp.77-84, ⟨10.1007/978-3-642-29461-7_9⟩. ⟨hal-00723989⟩