Evaluating product-based possibilistic networks learning algorithms

Nahla Ben Amor (1) , Maroua Haddad (1, 2) , Philippe Leray (2)

1 LARODEC - Laboratoire de Recherche Opérationnelle de Décision et de Contrôle de Processus
2 LINA - Laboratoire d'Informatique de Nantes Atlantique

This paper proposes a new evaluation strategy for product-based possibilistic networks learning algorithms. The proposed strategy is mainly based on sampling a possibilistic networks in order to construct an imprecise data set representative of their underlying joint distribution. Experimental results showing the efficiency of the proposed method in comparing existing possibilistic networks learning algorithms is also presented.

Domaines: Apprentissage [cs.LG]

Nahla Ben Amor, Maroua Haddad, Philippe Leray. Evaluating product-based possibilistic networks learning algorithms. 13th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty (ECSQARU 2015), 2015, Compiègne, France. ⟨10.1007/978-3-319-20807-7_28⟩. ⟨hal-01150813⟩