Possibilistic Classifiers for Uncertain Numerical

Authors: Myriam Bounhas Henri Prade Mathieu Serrurier Khaled Mellouli

Publish Date: 2011/6/29

Volume: , Issue: , Pages: 434-446

PDF Link

Abstract

In many realworld problems input data may be pervaded with uncertainty Naive possibilistic classifiers have been proposed as a counterpart to Bayesian classifiers to deal with classification tasks in presence of uncertainty Following this line here we extend possibilistic classifiers which have been recently adapted to numerical data in order to cope with uncertainty in data representation We consider two types of uncertainty i the uncertainty associated with the class in the training set which is modeled by a possibility distribution over class labels and ii the imprecision pervading attribute values in the testing set represented under the form of intervals for continuous data We first adapt the possibilistic classification model previously proposed for the certain case in order to accommodate the uncertainty about class labels Then we propose an extension principlebased algorithm to deal with imprecise attribute values The experiments reported show the interest of possibilistic classifiers for handling uncertainty in data In particular the probabilitytopossibility transformbased classifier shows a robust behavior when dealing with imperfect data

Keywords: