Interpreting Fuzzy Decision Trees with Probability-Possibility Mixtures
Abstract
The interpretation of the classification result obtained with a fuzzy decision tree is a not-so-easy task as the meaning of the obtained degrees of membership may depend on the type of fuzzy partition involved, ranging from probabilistic to possibilistic readings. Hybrid probabilistic-possibilistic mixtures can provide an interesting way to clarify the underlying components of such a classification result. In this paper, based on the hybrid-mixture model, a new approach is proposed to analyse the result of the classification with a fuzzy decision tree that enhance the explainability of this model.
Domains
Artificial Intelligence [cs.AI]Origin | Files produced by the author(s) |
---|